Series and Parallel System Reliability Calculator

Compare two-component series and active-parallel reliability, or estimate series survival from two normal component-life distributions.

Engineering screening tool—verify inputs and results independently. Use at your own risk.

Known-reliability inputs

Use when each component survival probability at the same service life is already known.

System reliability

Series reliability0.810000dimensionless probability

Active-parallel reliability0.990000

Interpretation0 = fail · 1 = survive

Series operation requires both components. Active parallel operation requires at least one component, with no common-cause or switching failure.

How to use

Select Component reliabilities for known probabilities or Normal component lives for mean/scatter data.
Enter both components’ required values in the shown units. Reliabilities must be 0–1; means and standard deviations must be positive; service life must be non-negative.
Select Calculate or edit a field to update the active result. Select Reset to restore the example inputs.
Read the canonical series probability and the parallel or component metrics. A larger probability means greater survival through the stated service life.
Correct any warning before use. Do not apply this model to dependent failures, repairable systems, standby switching, common-cause failure, non-normal life distributions, or more than two components without a justified extension.

Equations used

For statistically independent component events:

Rseries = R₁R₂
Rparallel = 1 − (1 − R₁)(1 − R₂)

For normal component lives, the implemented staged method is:

zᵢ = −round₃((t − μᵢ)/σᵢ)
φ(x) = exp(−x²/2)/√(2π)
Δz = 0.001; xⱼ = −5 + jΔz; kᵢ = 1000(zᵢ + 5)
Rᵢ,raw(t) = Σⱼ₌₁ᵏⁱ φ(xⱼ)Δz
Rᵢ(t) = round₃(Rᵢ,raw(t))
Rseries(t) = round₃(R₁,raw(t)R₂,raw(t))

round₃(x) = Math.round(1000x)/1000. Here μᵢ and σᵢ are mean and standard deviation of life in hours, t is non-negative service life in hours, zᵢ is the signed standardized survival margin, φ is the standard-normal density, and every R is dimensionless. The series result uses unrounded component sums before final rounding.

Worked example

Enter R₁ = 0.9 and R₂ = 0.9. Expected: Rseries = 0.810000, Rparallel = 0.990000. Parallel redundancy raises survival only when either component can perform the function.
In normal-life mode use μ₁ = 150 h, σ₁ = 30.5 h, μ₂ = 120 h, σ₂ = 10.2 h, and t = 100 h. Expected: z₁ = 1.639, z₂ = 1.961, R₁ = 0.949, R₂ = 0.975, and Rseries = 0.926.

Theory and method

A series system fails when either component fails, so independence turns the joint survival event into a product. An active-parallel system fails only when both components fail, producing one minus the product of their failure probabilities. The normal-life mode converts each mean and standard deviation to a survival margin at the common service life, integrates the standard-normal density using a 0.001 right-endpoint sum beginning at −5, and then applies the displayed staged rounding.

Assumptions and limits

Two statistically independent, non-repairable components with constant architecture.
Active parallel means either component alone can satisfy the function; switching, load sharing, and common-cause failures are omitted.
Normal-life mode assumes normal component lives and one common time unit. Normal models can imply negative life when scatter is large.
The finite density sum is retained for numerical consistency; it is not an exact modern normal CDF and can under-resolve extreme tails.
Results are screening probabilities, not safety factors or certification evidence.

References

Rausand, M. and Høyland, A., System Reliability Theory, 2nd ed., Wiley, 2004.
NIST/SEMATECH e-Handbook: Normal Distribution.

Related analysis: Continue in TriboSolver when component reliability inputs should be refined from a physics-based tribology simulation; retain an independent system-level reliability model for dependencies and common causes.