Engineering Tolerances and Fits

Why Does a Fraction of a Millimetre Matter So Much in Precision Engineering?

No manufactured component is ever perfectly the size its drawing specifies. Every machining process introduces variation. Cutting forces, thermal expansion, elastic deflection and the inherent capability limits of the machine all play a role. Engineering tolerances define how much of that variation is acceptable. Fits define what happens when two toleranced components come together in assembly.

Getting tolerances and fits right is one of the most consequential decisions a mechanical engineer makes. Specify them too loosely and the assembly fails to perform. Specify them too tightly and manufacturing costs escalate — sometimes dramatically — without delivering any additional engineering benefit. Therefore, tolerance and fit selection is not a clerical task. It is a genuine engineering discipline.

What a Tolerance Actually Defines

A tolerance defines the permissible variation in a dimension. It sets an upper and a lower limit. The actual manufactured size must fall within that range for the component to pass inspection. The difference between the upper and lower limits is the tolerance band.

A dimension specified as 25.000mm ± 0.025mm has a tolerance band of 0.050mm. Every component must measure between 24.975mm and 25.025mm to be acceptable. Furthermore, the tolerance band communicates directly to the manufacturer what precision level the machining process must achieve. A tight tolerance demands a more precise process, more careful tooling, more controlled conditions and more thorough inspection. Consequently, tolerance specification is one of the most direct levers an engineer has over manufacturing cost.

The Three Types of Fit – Clearance, Transition and Interference

When two mating components — typically a shaft and a hole — come together, the relationship between their dimensions defines the fit. Three fundamental fit types cover almost every engineering assembly requirement.

Clearance fits produce a gap between shaft and hole in all assembly conditions. The shaft is always smaller than the hole — even at its largest permissible size. Clearance fits allow relative movement between components. Engineers specify them for rotating shafts in plain bearings, sliding mechanisms and assemblies requiring regular disassembly. A running clearance fit for a rotating shaft in a bearing housing is a typical example. The clearance must be large enough to allow an oil film to develop. However, it must also be small enough to maintain shaft positioning accuracy.

Interference fits produce a condition where the shaft is always larger than the hole. Assembly requires force, heat or both. A heated outer component expands to accept the shaft. It then contracts as it cools — creating a permanent, high-friction joint. Interference fits transmit torque, resist axial loads and create rigid connections without fasteners. Gear hubs on shafts, bearing races in housings and wheel centres on axles are all interference fit applications. Furthermore, the magnitude of interference determines the clamping force and the load-transmitting capability of the joint. This makes interference fit calculation an important structural engineering task.

Transition fits occupy the middle ground. Depending on where within their tolerance ranges the actual manufactured dimensions fall, a transition fit assembly may produce either a small clearance or a small interference. Engineers specify them where location accuracy is critical but the joint must remain removable. Bearing fits where precise shaft location matters but routine maintenance requires removal are a typical example. Moreover, engineers must confirm that both the clearance and interference conditions are acceptable before specifying a transition fit — requiring careful tolerance analysis before the drawing is issued.

The ISO 286 System

Engineering fits in the metric system follow ISO 286 — the international standard for limits and fits. The system uses a letter-number designation to specify the tolerance zone for a hole or shaft. Capital letters denote holes. Lowercase letters denote shafts. The letter defines the position of the tolerance zone relative to the nominal dimension. The number defines the tolerance grade — the width of the tolerance band.

H7/g6 is a typical close running fit designation. H7 specifies the hole — fundamental deviation H, tolerance grade 7. g6 specifies the shaft — fundamental deviation g, tolerance grade 6. Together they define a clearance fit with a small but reliable gap suitable for precision running applications. Similarly, H7/p6 specifies a light interference fit. Engineers use it for a bearing race pressed into a housing where occasional disassembly may be required. Furthermore, H7/s6 specifies a heavier interference — suitable for a permanent press fit that must transmit significant loads without slipping.

ISO tolerance grades range from IT01 — extremely precise and costly — to IT18, which is very loose. Most engineering applications use grades IT6 through IT11. Moving from IT9 to IT7 can double machining time and triple inspection costs. Therefore, specifying the tightest available tolerance grade on every dimension is not conservative engineering. It is expensive engineering — and often unnecessarily so.

Tolerance Stacks — Why Individual Tolerances Are Not Enough

Individual component tolerances must be considered in combination, not in isolation. In any assembly involving multiple components, the tolerances accumulate. The variation in the overall assembly dimension is the sum of the variations in each contributing component. Engineers call this a tolerance stack.

Worst-case stack analysis calculates the maximum possible assembly variation by adding all individual tolerances. It guarantees that every assembly meets the requirement. However, it often produces very tight individual tolerances to achieve it. Statistical tolerance analysis takes a more realistic view. It recognises that components will not simultaneously sit at their worst-case limits. As a result, it produces wider individual tolerances while still achieving the required assembly performance at a defined confidence level.

Tolerance stack analysis is essential in any assembly with a critical overall dimension. A sealed mechanism with a defined end float, a gear train with a specified backlash requirement or an assembly with a minimum clearance requirement all need it. At CNR, tolerance analysis forms part of mechanical design and CAD development — ensuring assemblies meet their functional requirements across the full range of manufactured component variation.

Geometric Tolerances — Beyond Dimensional Variation

Dimensional tolerances control size. However, they do not control shape, orientation or position. A shaft diameter may be within tolerance while the shaft itself is bent. A hole may be the correct diameter but drilled off-centre. Geometric dimensioning and tolerancing — GD&T, governed by ISO 1101 in the metric system — addresses these additional dimensions of variation.

GD&T symbols on a drawing communicate requirements for straightness, flatness, circularity and cylindricity. They also cover perpendicularity, parallelism, angularity, position, concentricity and runout. Each controls a specific geometric characteristic independently of size. Furthermore, GD&T uses the concept of datums — reference features from which engineers measure geometric tolerances. This defines exactly how inspectors must orient and locate the part during measurement. Consequently, a drawing that correctly applies GD&T communicates the full geometric intent of a design — not just the size of each feature in isolation.

---

Tolerances, Cost and Engineering Judgement

The relationship between tolerance specification and manufacturing cost is one of the most important — and most frequently misunderstood — aspects of precision engineering design. Tighter tolerances always cost more. The cost increase is not linear. A step change in required precision — from a tolerance achievable by standard milling to one requiring grinding or hard turning — can multiply machining cost many times over.

Good tolerance specification requires the engineering judgement to distinguish between features where precision genuinely matters and features where it does not. A bearing bore requires a tight tolerance because bearing performance and life depend directly on it. A non-critical bracket hole does not require the same precision. Specifying it at the same grade adds cost without adding value. Moreover, over-tolerancing communicates poor engineering understanding to manufacturers — and erodes trust in the drawings that carry the specification.

At CNR, tolerance specification is an integrated part of detailed design and design for manufacture — grounded in 35 years of cross-sector experience across aerospace, automotive, defence, energy and research. If your programme needs tolerance and fit specification that balances precision with practicality, that engineering depth is where the conversation starts.