Tolerance stack-up
Builds a worst-case and statistical stack-up along a defined loop, ranks contributors by share of total variation, and tells you which two tolerances to tighten or which can be opened for cost.
HUMAN SKILL · IN THE AI ERA
For the engineer
Engineers are already using AI for mechanical design, but it's inaccurate. Datum is the world's first library of mechanical engineering skills - use them for your daily work or hand them to your AI agent, and share your skills to benefit other engineers worldwide. Read why we started Datum →
Top contributors
Browse all
Anders Holm
Daniel Weber
Hiroshi Tanaka
Julien Dubois
Kenji Aoyama
Lars Ekstrom
Marco Bellini
Marcus ReedShowing 12 of 12 sheets
Rev. Q3 2026
Builds a worst-case and statistical stack-up along a defined loop, ranks contributors by share of total variation, and tells you which two tolerances to tighten or which can be opened for cost.
Reviews a drawing's geometric tolerancing for legality and intent: datum precedence, degrees of freedom actually constrained, redundant or missing controls, and whether the callout can be inspected.
Sizes and verifies a bolted joint end to end: preload, stiffness ratio, separation and slip margins, fatigue on the alternating share, and a torque spec with the friction assumption written down.
Audits a simulation before anyone trusts it: reaction balance, hand-calc cross-check, singularity screen, convergence, and whether the reported stress is even the right quantity for the criterion.
Produces a documented first-principles calc sheet for beams, shafts, buckling, pressure, fits, springs, and bearings: assumptions numbered, units carried, margin reported against a named criterion.
Walks a machined part for manufacturability: setup count, internal radii, deep pockets, thin walls, workholding, and the tolerances that quietly force secondary operations and triple the price.
Normalizes supplier quotes to common scope and terms, builds a bottom-up should-cost from material, cycle time, and tooling, and identifies the specific line items worth negotiating.
Runs an Ashby-style selection: screens candidates on hard constraints, ranks on the right material index, states the failure mode each finalist is weakest against, and recommends grade and temper.
Facilitates a Design FMEA that teams actually use: function decomposition, physics-of-failure brainstorming, honest severity and occurrence scoring, and verifiable actions in RPN or AIAG-VDA format.
Drives a disciplined failure investigation: is/is-not problem definition, mechanism-first causal analysis, verified root cause with evidence, and a complete 8D covering technical, escape, and systemic causes.
Converts requirements into a verification matrix with pass/fail criteria, statistically based sample sizes with stated confidence, test sequencing, and the gaps where a requirement is untestable as written.
Drafts a complete engineering change order: itemized before/after, the interchangeability call, an exhaustive affected-items sweep, stock disposition, and the validation required before release.
Stay in the loop
Choose the disciplines you care about and we will send you every new skill in them. Work email only, so we can verify that our library stays a community of practising mechanical engineers.