Likelihood
In reliability engineering and CMMS, likelihood is the probability or frequency that a specific failure mode will occur during a defined operating period, often expressed as a conditional probability of failure at a specific age or as an annual failure rate. It is a core input for Risk Priority Number (RPN) calculations, where RPN = Severity × Occurrence (likelihood) × Detection.
In ServiceGrid or similar CMMS, likelihood is scored (e.g., 1–5) during Failure Mode and Effects Analysis (FMEA) to rank failure modes by criticality and prioritize preventive maintenance (PM) tasks. It drives maintenance optimization by adjusting PM intervals, changing tactics (e.g., from preventive to predictive), or redesigning assets to reduce the likelihood of failure. CMMS uses historical failure rate data to quantify likelihood, enabling automated MTBF/MTTR tracking and failure-mode distribution analysis.
- Underestimating likelihood by ignoring wear-out characteristics or conditional probability at specific ages, leading to catastrophic failures due to inadequate PM intervals.
- Using static likelihood scores without updating them with new CMMS failure data, causing recurring failures and wasted maintenance spend.
- Confusing likelihood with severity, resulting in misprioritized RPN scores and failure to address high-frequency, low-severity issues that degrade overall asset reliability.
How is likelihood different from reliability (R(t))?
Likelihood is the probability of failure (F(t)), while reliability is the probability of survival (R(t) = 1 − F(t)) over a mission time.
Can likelihood be calculated from MTBF?
Yes; failure rate (λ) ≈ 1/MTBF, and likelihood of failure in time t is F(t) = 1 − e^(−λt) for constant failure rates.
What data does ServiceGrid need to compute likelihood accurately?
Structured failure coding, run-time hours, work-order history, and criticality classification to feed FMEA and RPN models.
Why is conditional probability of failure more useful than simple failure rate?
It captures age-dependent wear-out (e.g., bearings failing after 2 years), whereas simple rate assumes constant failure probability, which misleads PM planning.
How does likelihood feed predictive maintenance (PdM) triggers?
High likelihood scores for degrading conditions (e.g., vibration spikes) trigger PdM alerts before failure, reducing downtime and MTTR.