How Tooth Spacing Error Shapes Load Sharing and Noise in Split Drives
Geared turbofan engines depend on split-path transmissions to divide torque across multiple parallel gear stages. When tooth spacing drifts away from its nominal value during manufacture, the contact pattern shifts, load sharing becomes uneven, and airborne noise from the gearbox rises. Understanding these effects matters for any program developing more efficient propulsion, whether the airframe is a regional turboprop flying out of Brisbane or a business jet taking off from Essendon Fields.
The OPTIMIZE Project studies tolerance analysis, design-of-experiments screening, and physical testing methods that allow engineers to anticipate how small geometric imperfections propagate into transmission error and acoustic radiation. The following sections walk through where spacing error originates, how it couples with bearing friction and hyperstatic loading, and what Australian aerospace teams can do to reduce its impact on next-generation power split designs.
Split-Path Architecture and Where Errors Enter the System
A split-path gearbox routes torque through two or more parallel gear branches that share a common carrier or output. In a typical epicyclic stage, the sun pinion meshes with several planets, and load is distributed across each planet-bearing pair. Manufacturing variations change the geometry of every tooth, every planet, and every bearing race, so the load never distributes itself as the textbook diagram suggests.
Spacing error is the deviation of the centre distance between adjacent teeth on a gear from its theoretical value, often expressed in micrometres. In a single-mesh pair, a few micrometres of spacing error produce a small static transmission error and a modest vibration tone. In a split-path drive, the same error multiplies because the carrier must accommodate the mismatch between branches, twisting the planet bearings and forcing one planet to carry more load than the designer intended.
| Configuration | Load Sharing Behaviour | Sensitivity to Spacing Error | Typical Acoustic Signature |
|---|---|---|---|
| Planetary epicyclic with three planets | Symmetric at design point | High; one weak planet overloads | Strong tonal content at mesh frequency and sidebands |
| Planetary with four or five planets | More uniform at design point | Moderate; redundancy helps | Broader tonal spread, lower peak amplitude |
| Split-torque bevel and spiral-bevel | Asymmetric by design | Moderate; each branch sets its own pattern | Lower mesh tones, higher broadband content |
| Counter-rotating open gearing | Equal in theory | Lowest; error absorbed across many teeth | Broadband, depends heavily on housing stiffness |
The columns above summarise trends observed in OPTIMIZE Project tolerance runs and in published rig data from European Clean Sky demonstrators. They are useful as a starting point, but actual behaviour depends on the bearing preload, the housing deformation, and the lubrication regime.
Sources of Tooth Spacing Error in Modern Manufacturing
Tooth spacing error enters a gearbox long before the gears reach the test bench. Hobbing, shaping, and grinding each leave their own signature on the involute profile and the circumferential pitch. Heat treatment adds distortion that grows during carburising or nitriding, while finish grinding only partly recovers the original geometry. Australian gear suppliers supporting local aerospace programs in Adelaide and Melbourne have reported that pitch errors of four to eight micrometres are common even on well-controlled production runs.
Beyond the gear itself, the supporting structure carries its own tolerances. Bearing seats on the planet carrier, bolt-hole patterns on the gearbox housing, and the concentricity of the ring gear mount all influence how the planets sit relative to the sun and ring gear. When two of these errors align, the planet that should carry twenty-five per cent of the load can end up carrying forty per cent. A useful indicator of such behaviour is the variation in planet-bearing reaction force measured during spin pit testing.
Run-in wear and plastic deformation also shift the effective centre distance during the first flight cycles. New gears that look perfect on a coordinate measuring machine can develop a small asymmetric contact patch after a few hours under load, especially when the lubricant film is thin. Engineers at the University of Melbourne and RMIT have studied these early-life changes as part of broader gearbox reliability research.
From Spacing Error to Transmission Error and Acoustic Radiation
Transmission error is the difference between the theoretical angular position of a driven gear and its actual angular position. Pitch error in the driver produces a periodic angular oscillation at the mesh frequency, which propagates through the carrier and the housing. The housing acts like a loudspeaker, and any tonal content within its acoustic range becomes a source of cabin noise or external flyover signature.
The link between spacing error and noise is not always obvious because the gearbox can absorb small errors through elastic deflection of the teeth, the bearings, and the housing structure. Once the elastic cushion is exhausted, however, the contact pattern moves to the edge of the tooth and the dynamic mesh stiffness changes abruptly. That change shows up as sidebands around the mesh tone and as high-frequency excitation that radiates efficiently through thin-walled sections of the casing.
Engineers chasing cabin-noise complaints in regional turboprops often trace the problem back to a single planet whose spacing error exceeded the build specification by just two or three micrometres. The fix can be as simple as selecting planets with matched pitch errors, or as involved as redesigning the planet bearing preload.
Coupling Spacing Error with Bearing Friction and Hyperstatic Loading
When a split-path gearbox is statically indeterminate, the system is described as hyperstatic. Adding more contact constraints reduces compliance and helps with stiffness, but it also means the gears cannot adjust themselves to small geometric errors. The carrier flexes, the bearings twist, and friction torques rise. A complementary study on bearing behaviour under combined high speed and external load is summarised at the friction torque findings page, and it shows how bearing response amplifies the effects of geometric error rather than absorbing it.
Hyperstatic loading also makes lubrication more difficult. A planet that carries too much load runs hot, the elastohydrodynamic film thins, and the surface traction coefficient climbs. That added traction shifts the dynamic mesh stiffness phase, which in turn modulates the transmission error waveform. The end result is a gearbox that is louder than predicted and harder to cool.
For Australian operators flying hot-and-high routes out of Mount Isa or Wyndham, any rise in gearbox loss translates directly into higher specific fuel consumption. Reducing spacing error is therefore not only a noise solution but also an efficiency lever that fits neatly into the OPTIMIZE Project mandate.
Sensitivity, Screening, and Statistical Design of Experiments
Because pitch error is statistical rather than deterministic, the engineering response must also be statistical. The OPTIMIZE Project team uses orthogonal array screening to identify which factors dominate the response and which can be left uncontrolled. A practical walk-through of the screening logic appears in the project write-up on Taguchi array notes.
The screening work typically includes planet-to-planet spacing error, ring gear runout, bearing preload, oil temperature, and housing stiffness as the most influential inputs. Once the dominant factors are known, tolerance budgets can be tightened on those features and relaxed on others, which lowers scrap rates without compromising acoustic performance. Practical moves to apply on a real program include the following.
- Match planet gears by measured pitch error before assembly rather than relying on part numbers alone.
- Specify tighter concentricity on bearing seats than the nominal drawing suggests, since this error compounds with planet pitch error.
- Map the housing deformation mode shapes and use the resulting stiffness matrix in transmission-error simulation.
- Run a spin pit acceptance test that records planet reaction forces, not just vibration spectra.
- Track lubricant film thickness during the test, because thin films magnify the contact-pattern shift caused by spacing error.
- Combine acoustic measurements in an anechoic chamber with structure-borne vibration on the test bench to separate housing radiation from gear meshing tones.
- Share datasets and quick-look plots with collaborating teams in Perth, Adelaide, and Sydney, using channels such as Telegram Plus for fast turnaround on plot reviews.
Practical Mitigations and How to Engage with the OPTIMIZE Project
For Australian aerospace teams, the cleanest path forward is to combine tolerance analysis, full-factorial or fractional-factorial screening, and a structured spin pit validation campaign. The OPTIMIZE Project provides videos, methodology notes, and a growing library of project publications that walk through each step of this workflow. The methodology is particularly relevant to local programs working on advanced rotorcraft drive systems, hybrid-electric powertrains, and small turbofan demonstrators that benefit from split-path architectures.
Readers who want to see how the research team connects design-of-experiments screening to physical testing can explore the project website and reach out through the contact channels listed there. Engineers, postgraduate researchers, and procurement teams across Australia can use the published case studies to set realistic tolerance targets, and they can contribute their own test data to broaden the statistical base. Bringing this knowledge into local programs will keep Australian propulsion work competitive with global Clean Sky demonstrators and will help the next generation of geared engines meet both efficiency and acoustic goals.