Shoulder fillet radius and bending stress in gear shafts
Geared aircraft engines place severe demands on the shafts that transfer torque between gears, bearings and power-reduction stages. Their stepped geometry creates abrupt changes in cross-section, while high rotational speed, fluctuating gear loads and fatigue cycles can turn small geometric discontinuities into crack-initiation sites. The shoulder fillet radius therefore has a direct influence on structural reliability and useful life.
The impact of gear shaft shoulder fillet radius on bending stress concentration extends beyond a single stress value. Designers must balance local stress, fatigue strength, shaft stiffness, mass, manufacturing cost and inspection access. An excessively sharp transition may initiate cracking, while an oversized radius can alter component packaging, increase inertia or move the critical stress into an unsuitable region.
The OPTIMIZE Project addresses this trade-off through an integrated approach involving design of experiments, numerical simulation, tolerance analysis and physical testing. This work supports the development of lighter, more durable and more power-dense gearboxes, particularly where every kilogram saved can improve aircraft payload, fuel efficiency or operating range.
From stepped geometry to local failure risk
A shaft shoulder connects a smaller diameter to a larger one, usually close to a gear, bearing or coupling. The material surrounding the transition must redirect longitudinal stress around the change in section. Without a generous transition, stress flow becomes distorted and a local peak develops on the smaller-diameter side of the shoulder. This is the geometric basis of the stress concentration factor, commonly written as (K_t).
For a solid circular shaft, the nominal bending stress at the smaller diameter can be estimated as (σ_{nom}=32M/(\pi d^3)), where (M) is the bending moment and (d) is the shaft diameter. The local elastic peak is then broadly represented by (K_t σ_{nom}). The factor depends strongly on the diameter ratio (D/d), the fillet radius-to-diameter ratio (r/d), shoulder geometry and the manner in which the load is applied.
A larger radius generally lowers (K_t) by spreading the change in section over a longer distance. However, it also removes more material from the shoulder and may require a larger machining blank. The designer must therefore identify the smallest radius that satisfies fatigue and strength requirements without compromising assembly, balance, bearing support or access for inspection and grinding.
Radius, fatigue life and operating limits
The geometric stress concentration is only one part of fatigue behaviour. Shafts experience repeated bending from gear forces, vibration, manoeuvring loads and changes in shaft alignment. Once a minute tensile stress concentration is present, the process of crack initiation becomes highly sensitive to surface condition, residual stress and local material properties.
The effective fatigue notch factor, (K_f), is often expressed as (K_f=1+q(K_t-1)), where the notch sensitivity (q) accounts for material ductility, notch sharpness and microstructural effects. A geometrically severe shoulder does not always produce a proportional reduction in fatigue life. Conversely, a modest reduction in calculated stress may be decisive when a highly stressed shaft is also polished, hardened or exposed to aggressive operating temperatures.
Fillet geometry also interacts with surface finish. Grinding can leave directional marks, while hard turning or peening may introduce beneficial compressive residual stress. A theoretical radius that is larger than the tool nose radius may still leave an effectively sharp surface feature. The relevant comparison is often between the specified fillet radius and the finished surface, rather than between nominal geometry alone.
High rotational speed adds centrifugal loading, balance sensitivity and runout concerns. Lubrication conditions influence thermal growth and the ability of oil to reach bearing and gear interfaces, while the shoulder itself can become a lubricant trap or flow disturbance. Effective shoulder design therefore considers bending concentration alongside torsional loading, keyway effects, bearing fits and the operating environment.
Manufacturing variation and toleranced outcomes
The best radius on a drawing is of little value if production cannot reproduce it consistently. CNC turning, form grinding and milling create finite tool radii, feed marks and transition-zone deviations. Undercutting may occur when the shoulder is machined close to the blank boundary, while an oversized radius can cause the finished feature to overlap a gear seat or interfere with an adjacent component.
Tolerance analysis should examine more than the mean radius. The important variables may include radial deviation, axial position, local surface waviness, shoulder runout and the position of the maximum bending moment relative to the transition. A statistically centred nominal value can still produce a field of fatigue results if the critical section coincides with the least favourable manufacturing outcome.
Design-of-experiments methods can connect these variables to responses such as peak von Mises stress, fatigue safety factor, shaft mass and manufacturing yield. A bounded design space can then be searched for robust combinations rather than a single theoretical optimum. Simulation-based sensitivity studies also reveal whether tightening the radius tolerance produces more benefit than improving surface finish or shifting the shoulder position.
Inspection plans should reflect the failure mechanism. Radius gauges, optical measurement and profilometry can verify geometry, while material records and process controls protect the surface condition assumed by the fatigue model. For flight-critical hardware, every actual part or representative test article should be traceable to the same geometry and process assumptions used in verification.
Simulation, testing and design experiments
Finite-element analysis provides detailed insight into the three-dimensional stress field around the shoulder. Axisymmetric models can support early comparisons of radius and diameter ratios, but a full shaft model is usually needed when gears, keyways, bearing fits or torque reactions break symmetry. The mesh at the fillet must be refined through a convergence study so that a further reduction in element size does not materially change the calculated stress concentration.
A linear-elastic model can identify where bending stress concentrates, although it does not by itself predict crack growth or fatigue life. Multiaxial fatigue criteria, surface corrections and appropriate material data are needed to translate the local stress into a life estimate. This is especially important in high-strength steels and titanium alloys, where the relationship between nominal stress, residual stress and crack initiation may differ from that of an ideal smooth specimen.
The gearbox must also be assessed under its actual load path. Hyperstatic conditions can distribute a gear force across several shafts and bearings, so changing one shoulder does not necessarily change system loading in isolation. Rotating-shaft analysis should include gear contact, bearing reactions, shaft deflection, torsional oscillation and the dynamic amplification associated with high operating speed. Engineers can also consult the mechanical design notes when developing or checking these geometric assumptions.
Physical testing is required to validate the simulation and the proposed manufacturing process. Instrumented fatigue specimens, rotating-bend tests and static proof loads can establish the effect of radius, surface finish and heat treatment. Strain measurement, digital image correlation and fractographic examination then show whether cracks initiate at the geometric peak or at a process defect. The resulting data can update the design-of-experiments model and provide evidence for the final allowable range.
Australian deployment, certification and supply chains
Australian aerospace programmes operate within a certification environment that makes traceability and conservative engineering evidence particularly important. Maintenance organisations approved under CASA Part 145 and design organisations approved under Part 21 have established responsibilities for approved designs, modifications and continued airworthiness. A gearbox shaft therefore benefits from a documented link between the fillet-radius limit, inspection method, fatigue evidence and released manufacturing instructions.
Local operating conditions also influence the commercial value of optimisation. Melbourne has a mature concentration of aerospace engineering, universities and advanced manufacturing organisations, making it a natural centre for collaboration on shaft durability and fatigue testing. Brisbane supports aerospace and precision-engineering development through its engineering community and research institutions. To the west, Perth’s mining sector provides extensive experience with high-load gearboxes and remote operations, although that industrial knowledge must be separated from aircraft certification requirements.
The Australian market rewards gearboxes that combine power density with maintainability, particularly for regional and charter operations where payload, fuel consumption and dispatch reliability affect the business case. Reducing shaft mass can benefit several aircraft systems at once, while extending fatigue life may reduce inspection frequency and replacement-stock requirements. These benefits must still be demonstrated against the certified operating envelope rather than inferred from mining or automotive applications.
Imported equipment is another practical consideration. Australian Customs processes may apply when a specialist grinder, radius gauge, inspection fixture or replacement shaft is purchased overseas, with tariff classification and supporting import documentation required before clearance. Local suppliers and workshops can reduce lead times for iterative testing, but procurement planning should include customs costs, calibration requirements and the time needed to qualify new manufacturing methods.
Turn the shoulder into a verified design choice
A robust fillet-radius decision begins with a consistent definition of the governing load case, shaft geometry and material condition. Engineers can then screen several radii using analytical factors and refined finite-element models before committing expensive physical testing. The preferred range should minimise sensitivity to manufacturing variation while preserving fatigue strength, stiffness, mass and assembly access.
Tolerance analysis should then connect nominal geometry to the as-built distribution of results. A practical approach treats the radius, surface finish, shoulder position and relevant heat treatment as linked variables, using design-of-experiments data to identify robust combinations. Validation articles can confirm that the chosen process reproduces the assumed geometry and that the measured stress agrees with the model.
The project documentation provides a practical reference for the project’s objectives and engineering methods. Review the existing methodology, nominate the shaft geometry and governing bending loads for the next study, and establish the radius, surface-finish and inspection limits that will form part of the gearbox verification plan.