How tooth root fillet roughness drives fatigue crack initiation
In modern geared turbofan engines, the reliability of every rotating component is governed by what happens at the millimetre scale. The fillet region at the root of a gear tooth is one of the most heavily loaded zones in the entire drivetrain, and the surface quality left behind by machining has a direct bearing on whether the part survives its design life. Even when the bulk geometry is sound, microscopic peaks and valleys along the root fillet can act as preferential sites for crack initiation, lowering fatigue strength in ways that bulk stress calculations alone cannot predict.
This question sits at the heart of recent research supported through the OPTIMIZE initiative, which targets measurable efficiency and durability gains for power reduction gearboxes in future aircraft engines. A study on the influence of gear tooth root fillet surface roughness on fatigue crack initiation looks specifically at how the texture of that curved transition zone interacts with cyclic loading. The findings carry weight for European Clean Sky demonstrators and for Australian companies that build, finish, certify and overhaul precision drivetrain hardware.
Researchers from Adelaide, Melbourne and Brisbane have collaborated with European partners to study how variations in surface finish, measured in micrometres of Ra and Rz, correlate with the number of cycles to first detectable cracking. The work draws on design-of-experiments methods, finite element analysis with realistic residual stress fields, and physical bench testing on instrumented gear specimens. Together they form a coherent picture of how a single finishing pass can shift the S-N curve by tens of thousands of cycles.
For Australian readers, the relevance stretches beyond aerospace primes. Operators based in Perth and Sydney maintain legacy fleets that demand stringent gearbox overhaul intervals, while emerging propulsion programs on the eastern seaboard are exploring regional supply chains for high-quality gearing. Understanding how roughness drives crack initiation helps these organisations make smarter decisions about procurement, process control and lifecycle planning.
Surface roughness mechanisms at the tooth root fillet
The fillet at the tooth root is geometrically a transition between two involute flanks and the bottom land. Its radius is deliberately chosen to spread bending stresses, but the curvature itself concentrates strain near the surface. When a hob or shaper cutter leaves behind a regular lay pattern of tool marks, the peaks of those marks become micro-notches. Under cyclic bending, each peak carries a localised stress that can exceed the nominal value by a factor that depends on the mark radius relative to the grain size of the steel.
Surface roughness is more than a cosmetic parameter. It interacts with residual stresses, work hardening, and the local microstructure introduced during case hardening or shot peening. A polished fillet that has been correctly superfinished can shift the fatigue endurance limit upward, while a rough surface produced by an aggressive roughing cut can introduce tensile residual stresses that accelerate crack nucleation. The roughness acts as a stress raiser, but the magnitude of that effect is strongly modulated by the manufacturing sequence that came before it.
Modern characterisation tools allow researchers to quantify these features using white-light interferometry and confocal microscopy, producing 3D maps of the fillet surface that feed directly into meshing finite element models. When those maps are loaded with realistic cyclic bending, the predicted locations of first plastic strain match closely with where physical specimens later reveal small cracks under dye penetrant or magnetic particle inspection. This convergence between simulation and bench testing is one of the cornerstones of the methodology explored within the broader project framework.
Stress concentration and crack initiation pathways
The classical picture of crack initiation at the tooth root places the highest tensile stress just beneath the surface, where the bending moment is greatest. Surface roughness modifies that picture. Sharp tool marks push the location of maximum stress even closer to the surface, into a region that is often metallurgically less refined due to decarburisation or grinding burn. The result is that cracks begin at lower applied loads and at shorter cycle counts.
Initiation can follow several pathways. In ground fillets, cracks often emerge at the bottom of grinding grooves and propagate along the direction of lay until they turn and grow into the bulk material. In superfinished or polished fillets, initiation occurs more frequently at microstructural features such as inclusions or grain boundaries, sometimes slightly below the original surface after a thin layer has been removed by the finishing process. The competition between these mechanisms explains why two gears with identical shape can behave quite differently under the same load spectrum.
Researchers are increasingly interested in how the direction of lay relative to the principal stress direction changes the effective stress concentration factor. Lay oriented perpendicular to the bending stress produces the worst-case scenario, while lay aligned with the stress axis produces a more benign response. Different gear cutting strategies produce different lay orientations, and post-processing steps like isotropic superfinishing can randomise the orientation and reduce the directional sensitivity.
Experimental methods and numerical correlation
Robust experimental work on how fillet surface texture drives fatigue crack initiation requires careful control of every variable except the one under investigation. Specimens are typically machined from a single bar of case-hardened steel, case depth and hardness are verified, and only the finishing operation at the fillet is varied. The finish is quantified using both 2D Ra values and 3D parameters such as Ssk and Sku, which capture skewness and kurtosis of the height distribution and correlate better with fatigue than Ra alone.
Fatigue testing is carried out on resonance or servo-hydraulic rigs, running at realistic gear pitch line speeds while applying controlled bending moments at the tooth root. Some programmes include pulsator tests on full gears run in back-to-back configurations, which add the effect of Hertzian contact stresses and lubricant film pressure to the bending fatigue picture. Crack initiation is detected using acoustic emission sensors, potential drop methods, and intermittent disassembly for inspection.
Numerical correlation relies on building finite element models from the measured surface profiles and applying the same boundary conditions as the rig. Elastic-plastic analysis with kinematic hardening, calibrated against monotonic and cyclic material data, captures the local stress-strain response at the rough peaks. The output of these simulations feeds into statistical lifetime models, often using the Dang Van or Sines criteria for multiaxial fatigue, producing probabilistic predictions of cycles to first cracking that can be compared directly with the experimental scatter bands.
Manufacturing variability in Australian aerospace supply chains
Australia hosts a sophisticated cluster of precision manufacturers serving both civil and military aerospace programs. Facilities around Melbourne, Brisbane, Adelaide and the Hunter Region produce gears, splines and gearbox housings for a range of platforms, often working under CASA airworthiness regulations and AS9100 quality systems. Within that cluster, finishing operations for root fillets vary widely: some shops rely on high-quality grinding with in-process gauging, while others depend on superfinishing or hand polishing to reach the required surface state.
Variability in finishing is one of the strongest hidden cost drivers for aerospace gearboxes. When surface roughness drifts above the specification, fatigue life erodes and the conservative design margin that engineers build into the part is silently consumed. Under the Work Health and Safety Act and CASA continuing airworthiness rules, an operator cannot simply increase inspection frequency to compensate; the root cause must be addressed through process control.
Regional defence and research bodies, including DSTG and several Group of Eight universities, have invested in concrete tooling for surface integrity studies. Their work supports local industry by providing independent measurement services, residual stress analysis using XRD, and access to fatigue rigs capable of running aerospace-relevant spectra. These assets are particularly valuable for smaller suppliers that lack the capital to build in-house capability.
Industry engagement with international programmes helps strengthen cohesion across the supply chain. Through collaboration with the European Clean Sky community, several Australian firms contribute to gear development projects that demand tight control of fillet surface integrity. The consortium participants page lists the organisations involved, including research centres and OEMs whose processes are benchmarked against best practice.
Implications for power reduction gearbox design
For power reduction gearboxes used in future geared turbofan architectures, every gram and every watt of loss matters. A reduction in allowable surface roughness at the tooth root can either permit a smaller fillet radius, reducing stress concentration without adding weight, or extend the gearbox's declared on-wing life between overhauls. Both outcomes support the higher power density targets set by the next generation of propulsion systems.
Designers can use the data emerging from roughness-focused fatigue studies to choose between competing surface treatments. Superfinishing, isotropic finishing, light shot peening and controlled grinding each leave a different combination of roughness, residual stress and microstructure. When the operating speed, lubrication regime and load spectrum are well defined, the trade-off becomes a defensible engineering choice rather than a guess.
The lessons also feed back into materials selection. Steels with cleaner inclusion populations, such as vacuum-arc remelted grades, tolerate rougher surfaces better than conventional steels because their largest inclusions are smaller than the surface notch root radii. This trade-off allows designers to balance the cost of premium steel against the cost of tighter finishing processes on standard grades.
Recommendations for engineers and procurement specialists
- Specify 3D surface parameters in addition to 2D Ra values when writing procurement documents, capturing skewness and kurtosis that better predict fatigue response.
- Build statistical tolerance stacks around surface roughness rather than relying on a single upper limit, so batch-to-batch variability is visible and controllable.
- Require suppliers to report the orientation of surface lay relative to the principal bending direction, since lay orientation can shift fatigue life by tens of percent.
- Include residual stress measurement as a release inspection when qualifying new finishing routes, especially for grinding operations that risk thermal damage.
- Validate finishing processes on representative test gears before approving them for production parts, using pulsator rigs that combine bending and contact loading.
- Partner with Australian research providers, including DSTG and university laboratories, to share measurement capability and reduce the cost of qualifying new processes.
- Track on-wing fatigue performance against finishing parameters in service, feeding field data back into design models to refine future specifications.
Every improvement in how we measure, control and verify the surface at the tooth root fillet translates into lighter, more durable gearboxes. The OPTIMIZE programme invites engineers, researchers and procurement professionals to explore the resources available through the broader project and to consider how these findings could reshape their own approach to gearbox design. If you are working on aerospace drivetrains in Australia or beyond, share this article with a colleague who wrestles with fatigue margins, surface finish specifications or supplier qualification, and reach out to the project team to start a conversation about collaboration.