Tolerance Stack-Up for Mechanical Engineers

3 min read

Tolerance stack-up analysis predicts how dimensional variation across multiple parts or features affects a functional requirement.

The useful workflow is:

function → dimension loop → contributors → calculation method → design decision

The formula is only one part.

When stack-ups matter

Use stack-up analysis when variation affects:

clearance;
interference;
alignment;
preload;
travel;
sealing;
bearing fit;
switch engagement;
cosmetic gap/flush.

If several dimensions influence the same function, the nominal CAD assembly is not enough.

Define the functional requirement

Start with what must work.

Example:

Final axial clearance must stay between 0.20 mm and 0.80 mm.

Do not begin by collecting every dimension on the drawing.

Build the dimension loop

Identify only contributors that change the functional output.

Assign sign based on direction.

A simple stack could be:

Clearance = housing depth − spacer thickness − bearing width.

The nominal result is only the center. Tolerance determines risk.

Worst-case method

Worst case assumes every contributor reaches its most unfavorable limit simultaneously.

For a linear stack:

Total worst-case variation is the sum of absolute contributor tolerances.

This is conservative and appropriate where interchangeability must be guaranteed under specified limits.

RSS or statistical approach

Root-sum-square methods assume variation behaves statistically and contributors are independent under suitable conditions.

For independent contributors:

RSS variation ≈ √(t1² + t2² + ...)

This can predict a narrower distribution than worst case, but the assumptions must be justified.

Do not use RSS merely to make the design pass.

GD&T contributors

Geometric tolerances can affect stack-ups differently from simple linear size tolerances.

Consider:

datum shift;
position;
orientation;
form;
material-condition modifiers.

Complex 3D stack-ups may require dedicated methods or software.

Worked example

Suppose:

housing depth = 20.00 ± 0.10 mm
spacer = 5.00 ± 0.05 mm
bearing width = 14.50 ± 0.05 mm

Nominal clearance:

20.00 − 5.00 − 14.50 = 0.50 mm

Worst-case total variation:

0.10 + 0.05 + 0.05 = 0.20 mm

So predicted clearance range is:

0.30 to 0.70 mm

If the functional requirement is 0.20–0.80 mm, the linear worst-case stack fits.

Now ask whether thermal expansion, geometric tolerances or assembly seating add contributors before declaring success.

Design fixes

If the stack fails, options include:

relax the functional requirement if justified;
tighten the dominant contributor;
change datum/interface strategy;
add adjustment/shim;
redesign to reduce contributors;
improve process capability;
choose matched/selective assembly only when justified.

Do not tighten every tolerance equally.

Documentation

Record:

functional requirement;
loop diagram;
contributors;
source drawing dimensions;
method;
assumptions;
result;
design action.

That makes the analysis reviewable.

Tolerance analysis is most valuable when it changes the design before variation becomes an assembly problem.

Sources

O*NET OnLine — Mechanical Engineers (17-2141.00), updated 2026 — Current occupation tasks, knowledge areas, transferable skills and technology categories.