Fit
Accumulated variation determines clearance, interference and whether mating parts can physically assemble.
Tolerance Stack-Up – Interactive Design Engineering Tool
A dimension is not just a number on a drawing — it represents permitted manufacturing variation. Build a dimension chain, compare worst-case and statistical accumulation, explore sensitivity, test functional limits and understand how datum strategy, GD&T and manufacturing capability influence real assembly performance.
Design → Tolerance → Variation → Function Worst Case · RSS · Monte Carlo Function first
20.000
Functional nominal
±0.300
Worst-case tolerance
±0.158
RSS tolerance
0.100
Worst-case min margin
PASS
Functional assessment
Why tolerance stack-up matters
Variation is inevitableAlignment
Small feature variations can combine into significant positional or angular error at the final interface.
Function
Sealing, bearing preload, gear mesh, connector engagement, optical position and many other functions depend on the resulting stack.
Quality & Cost
Overly tight tolerances raise manufacturing and inspection cost; loose or poorly allocated tolerances create escapes, rework and selective assembly.
WORST CASE
Assumes every contributor reaches its adverse limit simultaneously. Use when every assembly must function regardless of where parts lie within specification.
TWC = Σ |sᵢ|·Tᵢ
Assumes every contributor reaches its adverse limit simultaneously. Use when every assembly must function regardless of where parts lie within specification.
TWC = Σ |sᵢ|·Tᵢ
RSS
Combines independent contributors statistically using root-sum-square. Less conservative, but only defensible when process behaviour and assumptions justify it.
TRSS = √Σ(sᵢTᵢ)²
Combines independent contributors statistically using root-sum-square. Less conservative, but only defensible when process behaviour and assumptions justify it.
TRSS = √Σ(sᵢTᵢ)²
MONTE CARLO
Randomly samples assumed contributor distributions to estimate the resulting assembly distribution, yield and tail risk. Useful for larger or non-simple stacks.
Randomly samples assumed contributor distributions to estimate the resulting assembly distribution, yield and tail risk. Useful for larger or non-simple stacks.
Worked linear example
If X = A + B − C − D, with A = 20 ±0.10, B = 15 ±0.05, C = 10 ±0.10 and D = 5 ±0.05:
Nominal X = 20.000 mm · Worst Case = ±0.300 mm · RSS ≈ ±0.158 mm
The same nominal design therefore has different risk statements depending on the analysis method and assumptions.
Start with function, not tolerances
Function → Functional Requirement → Datum Strategy → Tolerance Chain → Manufacturing Capability → Measurement Capability → Allocation → Verification
The engineering question is not “how tight can we make every dimension?” It is “what variation can the function tolerate, and how should that allowable variation be allocated economically and measurably?”
GD&T changes the problem
Simple ± stacks are useful for one-dimensional chains, but real assemblies may also depend on position, orientation, form, datum feature variation, material-condition modifiers, bonus tolerance and datum shift. A linear arithmetic stack should not be used as a substitute for correct GD&T interpretation.
Common trap: adding coordinate tolerances while ignoring datum structure can overstate or understate the real assembly boundary. Establish the functional interface and degrees of freedom first.
Quick 4-Dimension Simulator
A + B − C − DLive result
20.000
Nominal result
±0.300
Worst-case
±0.158
RSS
19.700–20.300
Worst-case limits
Project Definition
Define the functional requirementFunctional requirement
20.000 mm
Calculated nominal
0.800 mm
Total permitted functional window
Dimension Chain
Sign controls direction in the stackEnter bilateral ± tolerances. Sign +1 adds the dimension to the functional result; −1 subtracts it. Sensitivity can scale a contributor if the relationship is not 1:1.
#DescriptionSignNominal± TolSens.σ basisProcess Cpk
20.000
Nominal
±0.300
Worst-case
±0.158
RSS
—
Monte Carlo 99.73% band
Stack Visualisation
PASSTolerance Contribution / Sensitivity
The largest contributors are normally the best places to investigate first. Tightening a very small contributor may add cost while barely changing assembly risk.
Monte Carlo Simulation
Educational statistical model—
Estimated yield inside functional limits
—
Simulated mean
Simulation assumes independent normal contributors centred on nominal unless otherwise stated. Real process shifts, truncation, correlation and non-normality can materially change risk.
Design Review Questions
- Is the functional requirement explicitly defined and physically meaningful?
- Is the stack path the shortest valid path between functional interfaces?
- Are datum references aligned with how the assembly locates in reality?
- Are thermal growth, coatings, adhesive bondlines, shims, wear or preload relevant?
- Does the manufacturing process demonstrate capability against the allocated tolerance?
- Is the measurement system capable of verifying the requirement?
- Would GD&T provide a clearer and more functional specification than coordinate ± dimensions?
- Has the analysis been validated against actual assembly data?
Relevant standards framework
- ASME Y14.5-2018 (R2024) — authoritative US GD&T language covering symbols, rules, definitions, defaults and recommended practices for dimensioning and tolerancing.
- ISO 1101:2017 — geometrical tolerancing language for form, orientation, location and run-out.
- ISO 8015:2011 — fundamental GPS concepts, principles and rules for specification and verification.
- ISO 14405-1:2025 — dimensional tolerancing requirements for linear sizes.
- ISO 5459:2024 — datums and datum systems.
- ISO 2692 — maximum/least material requirements and related GPS concepts where applicable.
Important: these standards establish specification language and interpretation. They do not replace product-specific functional analysis, customer requirements, company drafting standards, verification planning or competent design authority approval.
Design engineering best practice
- Define the functional interface first. State minimum/maximum clearance, gap, engagement, preload, position or other required outcome.
- Create a functional datum scheme. Constrain the same degrees of freedom that the real assembly constrains.
- Build the shortest valid stack. Avoid redundant dimensions and unnecessary chained dimensioning.
- Use baseline/datum dimensioning where appropriate. Chained coordinate dimensions can cause unintended cumulative tolerance.
- Allocate tolerance by sensitivity and capability. Give more tolerance to low-sensitivity contributors and avoid demanding precision from incapable processes.
- Use GD&T where function is geometric. Position, profile, runout, orientation and material-condition modifiers can communicate assembly intent better than ± coordinates.
- Include non-machined contributors. Coatings, plating, adhesive, gasket compression, thermal expansion, distortion and wear often dominate real-life stacks.
- Validate measurement capability. A tolerance that cannot be reliably measured is not an effective control.
- Close the loop with actual data. Compare stack predictions with process capability and assembly measurements; update the analysis as the design matures.
Avoid indiscriminate tightening
Tighter tolerances increase process, tooling, inspection and scrap cost. Tighten where sensitivity and functional margin justify it.
Avoid tolerance chaining
Where multiple features share a functional datum, baseline or GD&T control can reduce unintended accumulation compared with long chains.
Avoid “drawing-only” analysis
The real stack includes how parts locate, deform, heat, coat, bond, fasten and wear — not merely the printed nominal dimensions.
Improvement Guidance
Project Notes
Management Summary
Tolerance Stack-Up Analysis Report
Special Processes Institute · Design Engineering Analysis
1. Executive Assessment
2. Project & Functional Requirement
3. Key Results
Nominal
Worst case
RSS
Minimum WC margin
4. Dimension Chain
| # | Contributor | Sign | Nominal | ± Tol | Sensitivity | WC contribution | RSS contribution² |
|---|
5. Findings & Actions
Observations
Actions
Risks / Assumptions
6. Standards / Engineering Context
Consider the applicable product-definition framework, including ASME Y14.5 or ISO GPS standards such as ISO 1101, ISO 8015, ISO 14405-1 and ISO 5459, together with customer, contractual and organisational requirements. This report is an analytical aid and does not itself define drawing requirements.