Deterministic calculators for engineering, machining, quality and costing. See every input, assumption, formula and warning before you act. Client-side — your data never leaves your browser.
Different roles run the numbers for different reasons. Pick the question — we route you to the right calculator and the right evidence.
The hardest everyday question on the floor — will the assembled dimensions stay within spec? Three methods, one deterministic engine.
Worst-case for the safe bound, RSS for the statistical estimate, seeded Monte Carlo for the distribution — compared side by side, with the contributor that drives the most variation called out.
Every tool shares the same principles: structured inputs, full-precision math, visible assumptions, and a result you can defend.
Know the real number before you commit a price or a hire.
Predicted capability from tolerances — flagged as prediction.
See the contributor, the margin and the warning before it costs you.
A result is only useful if you can defend it. Every report carries four layers — so a reviewer can follow the number from input to decision.
Primary result, pass / warning / fail, and the operational or commercial impact — stated up front.
Every input with units, derived values, the formula used and the assumptions made. Nothing hidden.
Input consistency checks, outlier warnings, the applicable standard and edition, and the engine version.
Executive result, detailed calculation, deviations, deterministic hashes and a PDF / shareable record.
Not a black-box number. A traceable record — the kind you can hand to a reviewer or attach to a design file.
We don't claim to replace measured data or a licensed engineer. We claim to make the calculation transparent, repeatable and reviewable.
Structured, unit-aware inputs with tooltips that state the assumption — no hidden defaults.
The method and the math are stated in the report, so a reviewer can follow the derivation.
Out-of-range inputs, thin margins and out-of-spec predictions are called out, not buried.
Deterministic seed + engine version + hashes mean the same inputs reproduce the same report.
Trust comes from stating limits as clearly as capabilities.
1D linear stacks only. Per-part sigma is derived from the tolerance (no measured-process input yet). No correlation model between dimensions yet. Results are predicted from drawing tolerances — an engineering preview to support a decision, not a substitute for measured SPC or a licensed engineer's sign-off, and not for production approval on their own.
No subscription. Try every tool free first; open a premium calculation with credits valid for 12 months. Heavy users save with big packs; light users pay little.
See credit pricingRun a tolerance stack-up in your browser — no login, no upload, reproducible by seed.
Run a Calculation