Tolerance Analysis — Stack-Up and Worst Case

Tolerance analysis predicts the variation in a critical assembly characteristic (the "gap" or "closing dimension") from the combined effect of tolerances on all contributing features. The two primary methods are Worst Case Analysis (WCA) — guarantees 100% assembly if all individual tolerances are held — and Root Sum Square (RSS) — a statistical approach that accepts a small fraction of assemblies outside limit.

Why companies use it

·Prevents the design and redesign cycle caused by discovering that parts do not assemble due to accumulated tolerances
·Provides a rational basis for tolerance assignment — tighter tolerances cost more, so only tighten where the stack-up analysis shows it is needed
·Required in automotive (APQP), aerospace (AS9100), and medical device design control processes
·Enables dialogue between design and manufacturing about where to invest tolerance-tightening effort and where to relax

What hiring managers look for

·Mechanical engineers who cannot perform a tolerance stack-up analysis frequently deliver designs that cannot be assembled or measured
·Understanding when to use WCA (safety-critical, low-volume) versus RSS (high-volume, statistical acceptance) shows design judgment
·Experience with 1D, 2D, and 3D tolerance analysis tools (CETOL, Sigmund, DCS, Vis-VSA) is directly practical
·Closing the loop between tolerance analysis and manufacturing capability (Cpk) shows systems engineering thinking

Typical interview questions

What is the difference between a Worst Case Analysis and a Root Sum Square tolerance analysis?

You have a gap between two assembled components that must be between 0.1 mm and 0.5 mm. How do you determine the individual feature tolerances?

When does a RSS analysis give significantly better tolerance than WCA, and why?

How do you incorporate process capability (Cpk) data into a statistical tolerance analysis?

Describe a tolerance stack-up problem you have solved on a real design.

Common mistakes

·Using RSS analysis for safety-critical gaps without understanding that RSS allows a statistically predicted number of out-of-specification assemblies
·Performing WCA and concluding that tolerances must be unrealistically tight, without first applying the bonus tolerance available from GD&T MMC modifiers
·Including tolerances in the stack-up that are not actually in the assembly loop — only tolerances that contribute to the closing dimension belong in the analysis
·Not validating the tolerance analysis against measured data from actual production — the model is only as good as the input tolerance assumptions
·Performing 1D stack-up for a 2D or 3D geometric relationship — this gives incorrect results when contributions are not aligned in one axis

Real engineering example

A precision camera module assembly was experiencing 8% out-of-focus yield at final inspection. A 1D tolerance stack-up of the optical path length revealed that the combined worst-case tolerance was ±0.18 mm, but the focus depth of field required ±0.10 mm. By adding an active alignment step (measuring focus during assembly and adjusting with a shim), the process tolerance was reduced to ±0.03 mm and yield improved to 99.2%.

Topics covered

tolerance analysisstack-upworst caseRSSGD&Tassemblymechanical engineering

Related interview guides

Mechanical Engineer →Manufacturing Engineer →Mechatronics Engineer →Optical Engineer →