InfyCalculator
⚛️

Tolerance Stack-Up Calculator (Worst Case & RSS)

Assembly gap or overall dimension from a chain of toleranced parts — worst-case and statistical (RSS) stack-ups, the resulting min/max, how much tolerance you could relax under RSS, the sigma level of a required gap, and per-part contributions.

Dimensions in the loop — nominal±tolerance, with a leading − for dimensions that reduce the gap (one per line or comma-separated)
Required minimum result (e.g. clearance must be ≥)
Required maximum result
Process capability assumed for RSS (tolerance = ±k σ)
Loading calculator…

How it works

Nominal = Σ (±nominal); worst case = Σ tolerances; RSS = √Σ tolerance²; assembly σ = RSS ÷ k (tolerances taken as ±kσ); reject rate = P(result < min) + P(result > max) from the normal distribution

When parts stack up in an assembly, their tolerances stack too. The worst-case method adds every tolerance and guarantees the gap even if each part sits at its limit in the worst direction — safe, but it forces tight, expensive tolerances on long chains. The statistical (root-sum-square) method notes that parts rarely all land at their limits at once; if each dimension is normally distributed and independent, the assembly variation is the square root of the sum of squared tolerances, typically 30–50% less than worst case, letting individual tolerances loosen for the same assembly yield at the cost of a small, calculable reject rate. Which to use is a business decision: worst case for safety-critical or low-volume fits, RSS for mass production with capable processes. Sign each dimension by whether it opens or closes the gap, and check the nominal sits inside the window before worrying about tolerances.

Worked example

Three parts (+50 ±0.10, +25 ±0.05, +12.5 ±0.05) inside a housing (−87 ±0.15) leave a nominal 0.5 mm gap. Worst case is ±0.35 (0.15 to 0.85 — within a 0.1–0.9 requirement); RSS is ±0.194 at 3σ, and the predicted reject rate is about 0.2 ppm, so the tolerances could be relaxed roughly 2× under RSS while still meeting the window statistically.

Frequently asked questions

Why does RSS underestimate scrap on my line?

Because real processes are not centred (mean shifts) or not normal. Bender’s method (1.5 × RSS) or Monte Carlo with measured distributions gives more realistic figures.

How do I include a ±0.05 on a diameter that affects a radial gap?

Use half the diametral tolerance (±0.025) in a radial stack; convert every dimension to the direction of the gap before entering it.

Related calculators

Shop reference values; adjust for tool maker recommendations, machine rigidity, coolant and the actual material condition, and verify fits and tolerances against the applicable standard (ISO 286 / ASME B4.1, ISO 965 / ASME B1.1).