Tolerance stack up is the cumulative effect of individual part tolerances when multiple components come together in an assembly. Every manufactured part has small allowable variations in its dimensions. When you line up several parts, those small variations add up, and the total accumulated variation can be much larger than any single part’s tolerance alone. Tolerance stack up analysis calculates the maximum and minimum distance between two features in that assembly, telling you whether your parts will reliably fit together.
Why Individual Tolerances Add Up
No part is manufactured to its exact nominal dimension. A shaft meant to be 10 mm might actually be 10.02 mm or 9.98 mm, and both are acceptable if the tolerance allows ±0.05 mm. That’s fine for one part in isolation. But when you bolt five or ten parts together in a line, each one can be slightly too long or slightly too short. The total variation across that chain of dimensions is the stack up.
Think of it like stacking books on a shelf. If each book’s thickness can vary by 1 mm, and you stack 10 books, the total height could vary by as much as 10 mm. That 10 mm swing might mean your books no longer fit on the shelf. The same principle applies in mechanical assemblies: a gap that should exist between two parts might vanish entirely, or a part might not seat properly, because every dimension in the chain drifted in the same direction.
The chain of dimensions from one feature to another is called the “stack path.” Engineers trace this path through every part and gap in the assembly, identifying each dimension and its tolerance. The result tells them whether the final clearance or fit will stay within acceptable limits.
Worst-Case Analysis
The most conservative approach assumes every part simultaneously hits the extreme end of its tolerance range. If every dimension that adds to the total is at its maximum, and every dimension that subtracts is at its minimum, you get the largest possible assembly variation. The formula is straightforward: add up all the individual tolerances. For a stack of parts with tolerances T₁, T₂, T₃, and so on, the total worst-case tolerance equals T₁ + T₂ + T₃ + … + Tₙ.
This guarantees that 100% of assemblies will work, provided every individual part actually meets its tolerance. That’s the upside. The downside is that it’s overly conservative. In reality, it’s extremely unlikely that every single part drifts to the same extreme at the same time. The result is that worst-case analysis often forces designers to specify tighter (and more expensive) tolerances on individual parts than they truly need.
Statistical (RSS) Analysis
The statistical approach, commonly called Root Sum Squared or RSS, takes a more realistic view. Instead of assuming every part is at its worst, it assumes part dimensions follow a bell curve distribution centered on the nominal value. Most parts cluster near the middle of their tolerance range, and relatively few land at the extremes.
Under this assumption, the total assembly tolerance is the square root of the sum of each individual tolerance squared. For 10 parts each with a tolerance of ±0.1 mm, worst-case analysis gives you ±1.0 mm total. RSS gives you √(10 × 0.1²) = ±0.316 mm, roughly a third of the worst-case number. The tradeoff is that RSS doesn’t guarantee every single assembly will work. It typically covers 99.73% of assemblies (corresponding to three standard deviations), meaning about 3 in 1,000 assemblies could fall outside the predicted range.
A comparison from research on prefabricated construction illustrates the difference clearly. For one assembly, the actual measured deviation was about 11 mm. The worst-case method predicted 19.8 mm (far too conservative), RSS predicted 4.6 mm (too optimistic), and a Monte Carlo simulation landed at 15.4 mm, closer to reality while still providing a safety margin.
Monte Carlo Simulation
Monte Carlo simulation takes a different approach entirely. Instead of using a single formula, it runs thousands or millions of virtual assemblies. Each run randomly selects a dimension for every part based on its expected distribution, then calculates the resulting assembly dimension. After enough runs, you get a probability distribution of outcomes that shows not just the range but how likely each outcome is.
This method handles complexity that formulas struggle with. Traditional worst-case and RSS methods work well for simple linear chains of dimensions (1D problems), but real assemblies often involve angles, curved surfaces, and parts that interact in three dimensions. Monte Carlo simulation can model these complex geometric interactions directly, especially when paired with 3D CAD software. It also lets you use realistic, non-normal distributions for part dimensions, which matters when a manufacturing process tends to skew parts toward one end of the tolerance band.
1D, 2D, and 3D Stack Ups
The simplest stack ups are one-dimensional: parts stacked in a single direction, like discs on a shaft. A spreadsheet handles these well. You list each dimension and tolerance in the chain, apply worst-case or RSS, and get your answer.
Two-dimensional analysis becomes necessary when designs include angles, cams, levers, or features that aren’t collinear. A tolerance on an angular feature, for example, creates variation in two directions simultaneously. 2D analysis can also evaluate functional requirements beyond simple fit, such as forces, deflections, and kinematic behavior. This level typically requires dedicated software with geometric solvers that go well beyond what spreadsheets can do, and it’s often used during conceptual design to explore how different tolerance choices affect performance.
Three-dimensional analysis is the most complex and is usually reserved for late-stage design validation. It works directly from 3D CAD models and catches fit-related problems that 1D or 2D analysis might miss, such as interference between complex curved surfaces. The software requires advanced training and significantly more effort to set up, so it’s typically used as a final check rather than an everyday design tool.
A Practical Example: Hole Alignment
One of the most common real-world stack up problems is aligning holes across multiple parts. In aerospace manufacturing, for instance, fasteners need to pass through matching holes in skins, stringers, and frame panels simultaneously. Each hole has a positional tolerance that allows its center to drift slightly from the nominal location. When you try to pin two or three parts together, the positional errors in each part stack up, and a fastener that should slide through cleanly might not fit.
Analysis of this problem shows that for two parts with holes at the same nominal location, the clearance between the fastener and the hole must account for the RSS of both parts’ positional tolerances. As the number of hole pairs increases (say, a row of 20 fasteners along a seam), the odds that at least one pair will be tight enough to cause problems go up. The analysis adjusts for this, requiring slightly more clearance for patterns with many holes to maintain the same overall success rate.
How Stack Up Analysis Affects Cost
There’s a direct and well-documented relationship between tolerance and manufacturing cost: the tighter the tolerance, the more expensive the part. Tighter tolerances may require slower machining, more precise equipment, additional inspection steps, or higher scrap rates when parts fall outside spec. The rule of thumb is simple, but the cost curve is steep. Cutting a tolerance in half rarely doubles the cost; it can multiply it several times over.
Stack up analysis gives designers the information they need to allocate tolerances intelligently. Instead of making every part in the assembly equally precise, you can identify which dimensions in the stack path contribute most to the final variation. A dimension with a large sensitivity coefficient (it has an outsized effect on the final gap) deserves a tighter tolerance. A dimension that barely affects the outcome can be loosened, saving manufacturing cost with no meaningful impact on assembly quality.
This optimization process balances manufacturing cost against quality metrics like yield (the percentage of assemblies that work without rework). Tolerance-cost models account for machining costs, tooling, inspection, scrap, rework, and rejection rates, all as functions of the assigned tolerances. Research in prefabricated construction found that selecting alternate fabrication processes based on Monte Carlo tolerance analysis reduced rework risk by over 65%, a significant cost savings that comes directly from understanding how individual tolerances contribute to the final assembly.
The Role of GD&T Standards
Tolerance stack up analysis relies on a shared language for defining and communicating tolerances. In the United States, that language is governed by ASME Y14.5, the standard for geometric dimensioning and tolerancing (GD&T). The current version, Y14.5-2018, was reaffirmed in 2024 and establishes the symbols, rules, and definitions engineers use to specify not just size tolerances but geometric controls like flatness, perpendicularity, parallelism, and position.
These geometric tolerances often matter more than simple size tolerances in stack up analysis. A part can be the right length but tilted slightly, and that tilt creates variation in the assembly that a size-only analysis would miss. Properly applying GD&T ensures that the tolerance values used in a stack up actually reflect how the parts will behave when assembled, making the analysis meaningful rather than just a mathematical exercise.