Shaft surface finish and seal performance in high-speed gearboxes
Aerospace propulsion has entered an era where every gram, every watt, and every drop of lubricant is scrutinised. The OPTIMIZE Project, a Clean Sky-driven research effort studying power reduction gearboxes for geared turbofan engines, places seal behaviour at the centre of its efficiency story. A geared engine architecture relies on transferring torque through compact epicyclic stages spinning well beyond 10,000 rpm, where the smallest departure from a perfect interface surfaces as heat, vibration, or oil loss. Within Australia, where Qantas's domestic fleet and the Royal Australian Air Force's P-8A Poseidons demand both fuel efficiency and dependability over long transcontinental sectors, the cost of a sealing defect multiplies quickly across thousands of flight hours.
Most published attention on gearbox losses focuses on bearing drag, churning, and windage, yet seal leakage quietly erodes performance and reliability in parallel. The texture left behind on a shaft after grinding, superfinishing, or polishing dictates how the seal lip engages, how it wears, and ultimately how much oil migrates past it. Getting that texture right depends on more than choosing a low Ra number — it requires understanding how machining direction, lubricant chemistry, thermal growth, and operating speed all converge at a contact patch measured in micrometres.
The mechanics of seal-shaft contact
At the heart of any rotating seal interface sits a counterface machined into the shaft and an elastomer lip that rides against it. The two never touch over a continuous plane; instead, asperities on the shaft meet the rubber in a stochastic dance governed by local pressure, sliding speed, and thermal history. Radial lip seals in particular are designed to operate within a narrow band of radial load. Too little contact and oil slips past unchecked; too much and frictional heat builds until the elastomer hardens, cracks, or glazes over.
In geared aircraft transmissions running at 12,000 to 20,000 rpm, that contact band is under continuous centrifugal pumping action. Oil dragged along by the shaft is flung outward, but a fraction always wants to creep inward along the seal-to-shaft boundary, drawn by capillary action into the valleys of even a well-finished journal. Even so-called mirror finishes retain asperities that act as miniature reservoirs and abrasive ridges, eating away the seal lip one micro-abrasion at a time. Engineers at DSTG's Melbourne laboratories have observed seal failures traced to surface textures that passed visual inspection but failed profilometry, where Ra measured 0.4 µm rather than the specified 0.2 µm.
Hyperstatic loading — the condition created when multiple bearings constrain a shaft so tightly that thermal growth cannot relieve without internal stress — further amplifies seal distress. A shaft that deflects microscopically under load shifts the contact band on the seal, scrubbing wear into a narrower ring rather than spreading it evenly. Over thousands of hours, that scrubbing turns a controlled wear zone into a leakage path.
Quantifying surface finish beyond Ra
Workshop drawings still default to specifying a single roughness average, Ra, but Ra alone is a remarkably poor predictor of seal life. Two surfaces ground to the same Ra can behave wildly differently because of their peak count, valley depth, and the lay pattern left by the tool. The Rz parameter, defined as the average of the five highest peak-to-valley distances in a sampling length, captures what an elastomer lip actually feels — the rare-but-deep features that scrub hardest.
Material ratio curves, often denoted Rmr or Rk parameters, describe how much of the surface consists of peaks versus valleys at given depths. A bearing-friendly surface aims for a high plateau with small valleys, optimising lubrication retention while keeping the contact area solid. Seal-friendly surfaces, in some cases, are tuned slightly differently: too glossy a plateau can starve the lip of hydrodynamic fluid film, causing dry contact and accelerated wear. Australian standards referenced within the project include the AS 1380 series on surface finish measurement, with calibration houses like the National Measurement Institute in Lindfield providing traceability back to international length standards.
Lay direction matters as well. A circumferential grind lay — produced by a cylindrical grinding pass with fine-feed dressing — runs parallel to the seal lip's motion and reduces the chance of a scratch propagating circumferentially. Crosshatched or axial lay patterns, by contrast, create chatter marks that scoop oil back along the lay channels. Selecting the right lay, in the OPTIMIZE methodology, is treated as a deliberate design parameter rather than an artefact of the machine tool.
Speed, heat, and pressure at the seal interface
Centrifugal force dominates everything else in a high-RPM seal cavity. As shaft speed rises, oil clinging to the rotating surface is flung outward, lowering the partial pressure at the seal ID and pulling additional oil inward to fill the void. This pumping effect is proportional to roughly the square of speed, which is why even a tiny leakage path that looks trivial at 3,000 rpm can become a major oil loss above 15,000 rpm.
Thermal expansion compounds the problem. A steel shaft grows faster than an aluminium housing under the same temperature rise, so the seal gap closes as the gearbox warms up. In a typical power reduction gearbox, sun and planet gears dump heat into the lubricant, raising bulk oil temperatures past 120 °C during a climb segment out of Brisbane or Sydney on a hot summer afternoon. The seal elastomer, often a fluoroelastomer such as FKM or a more modern perfluoroether polymer, must hold hardness and resilience over that temperature band — and any drift in hardness shifts the contact pressure profile directly.
Internal pressure differentials across the seal also drive leakage. Scavenge pump design, gear-mesh blow-by, and aeration of the oil all build a positive pressure on the inboard side of the seal. When that pressure exceeds the seal's ability to resist, even a tiny texture-driven gap opens up and oil escapes. The OPTIMIZE team has measured cyclic pressure transients in test rigs at its partner facilities near Fishermans Bend that correlate directly with both leakage rate and seal-lip temperature.
Lubricant chemistry and material pairings
The choice of seal elastomer is meaningless without considering the lubricant it lives in. Modern aerospace gear oils carry extreme-pressure additives, anti-wear packages, and sometimes trace amounts of ester base stocks that swell certain elastomers but harden others. Harsh additive chemistries can etch the shaft surface itself over time, slowly turning a 0.1 µm finish into something closer to 0.3 µm. A seal designed for a specific Ra at commissioning can find itself running against a rougher counterface halfway through a TBO cycle.
Australian operating conditions add another layer. Aircraft based in Townsville or Darwin cycle through humid, salt-laden air that finds its way past air seals and corrodes exposed shaft journals during ground time. Tropical moisture trapped under a protective boot can flash to steam during a cold soak at altitude, leaving light oxidation pits that change local surface chemistry. Down south in Avalon or Edinburgh, winter mornings bring the opposite challenge: cold starts push lubricant viscosity sky-high, starving the seal of hydrodynamic film until oil temperature catches up.
Material pairing also affects friction. A seal running against a nitrided or induction-hardened journal sees different frictional heating than one running against a thin-film coating like DLC or CrN. The OPTIMIZE Project has explored these pairings as part of its tolerance analysis, since a coating that improves wear resistance can also alter thermal conductivity at the contact patch — and therefore influence how aggressively the elastomer ages during transient thermal cycles.
Bringing it together through testing and modelling
No amount of theoretical analysis substitutes for measurement on a representative test rig. The OPTIMIZE methodology leans heavily on design-of-experiments matrices that sweep surface roughness, lay direction, lubricant type, seal compound, and operating speed across their realistic ranges. Each run logs leakage rate, seal-lip temperature, frictional torque, and wear debris counts. Response surface models built from these data let engineers predict seal life and oil loss for any combination inside the design envelope, including combinations not directly tested.
Australian partners contribute specialised capability here. Local test houses in Melbourne and Brisbane operate aerated lubricant rigs at speeds exceeding 25,000 rpm, with on-line particle counters and oil sampling taps that feed back into the project's data lake. A full description of the test protocols, DOE structures, and tolerance analysis methods is available through the project's technical documentation, where practitioners can review the assumptions and reference measurements behind the published results.
Simulation complements physical testing by isolating individual parameters that would be impractical to vary one at a time on a real rig. CFD captures the churning and windage losses that contribute to seal-cavity pressure, while thermal models predict the temperature gradient across the seal lip. When seal wear models couple back into a tolerance analysis of the entire gearbox stack — bearing clearances, gear geometry, housing bores — the small decision about whether to specify Ra 0.2 or Ra 0.4 µm reveals itself as a measurable lever on efficiency, durability, and oil consumption.
Sealing performance in a high-speed geared transmission rests on countless small decisions made along the manufacturing chain. Shaft surface finish sits alongside bearing selection, lubricant choice, and seal compound as a knob that, once turned, has consequences across the whole system. The OPTIMIZE Project continues to refine how those knobs interact, helping Australian and European aerospace partners squeeze out the leakage paths that cost fuel, oil, and reliability.
Teams keen to collaborate, contribute, or follow this work can contact the project team to discuss testing slots, data-sharing arrangements, or follow-on research proposals aligned with the wider Clean Sky agenda.