Designing a labyrinth seal for high-speed shafts
Oil escaping from a high-speed gearbox shaft is more than a housekeeping issue. A thin film can travel along the rotating surface, become entrained in the airflow and form an oil mist near hot components. In an aircraft propulsion system, that loss of lubricant also threatens component cleanliness, worker exposure and the visibility needed to inspect a maintenance area. To prevent oil leakage around high-shaft speeds, a labyrinth seal replaces uncontrolled escape with a repeatable pressure barrier.
The OPTIMIZE Project provides a fitting framework for this challenge because it applies design-of-experiments methods, simulation, tolerance analysis and physical testing to power-reduction gearboxes for geared aircraft engines. Those tools work together: simulation narrows the design space, statistics reveal which variables matter, tolerances expose fragile assemblies and a test rig checks whether predicted leakage and thermal behaviour survive real conditions. The result should be a seal whose performance can be explained, rather than a prototype that happens to pass one run.
A labyrinth seal replaces rubbing contact with a controlled fluid path. Straight-cut teeth on a rotating shaft interleave with matching teeth in a stationary housing, producing a series of small pressure drops. A spiral or axial arrangement can create a more directional barrier. The goal is controlled flow: restrict external leakage while managing the oil that must pass through the seal, control windage and prevent pressure from building in the bearing housing.
For Australian aviation and advanced manufacturing, the payoff depends on maintainable, repeatable performance. Regional airlines, charter operators, Australian Defence, agricultural aviation and mining services operate across long distances where an unscheduled gearbox inspection can be costly. Hot conditions around Perth, the Pilbara and the Northern Territory alter oil viscosity, while salt-laden coastal air around Sydney and Melbourne raises corrosion and surface-quality concerns. Australian deployment must also respect CASA-approved design and maintenance controls, making predictable sealing performance especially valuable.
Why leakage becomes a design problem
Leakage is driven by pressure difference, centrifugal action, aerodynamic pumping and the condition of the interface. Shaft runout can periodically close a clearance, thermal growth can open it, and a rough or misaligned surface can create a pumping path. Oil viscosity changes the balance between pressure-driven flow and windage. This is why a seal selected for steady laboratory conditions may behave differently after thermal cycling, assembly tolerance stack-up or service wear.
Three principal approaches provide a useful baseline for design reviews. Their differences affect leakage, heat generation, sensitivity to manufacturing variation and suitability for a high-speed shaft. The comparison should be treated as an early screening choice, followed by analysis and testing for the actual gearbox.
| Characteristic | Contact lip seal | Radial labyrinth seal | Axial or spiral labyrinth seal |
|---|---|---|---|
| Contact action | An elastomer lip rubs against the shaft | Interleaving shaft and housing teeth form repeated restrictions | Axial or helical passages create a directional pressure barrier |
| Leakage control | Depends on compression, surface finish and lubricant wetting | Relies on clearance, tooth geometry and pressure balance | Uses length, helix direction and clearance to oppose flow |
| High-speed behaviour | Compact and effective, but friction and heat can rise quickly | No rubbing contact; windage and tooth impact still require attention | Low contact burden, though pumping effects can be sensitive |
| Principal risks | Lip wear, chatter, hardening and pressure accumulation | Runout sensitivity, burrs, thermal growth and incorrect venting | Reverse pumping, installation error and difficult cleaning |
| Best fit | Compact housings where wiping action and simple retention are priorities | High-speed rotating interfaces needing a non-contact barrier | Systems with space for directional flow control |
Contact lip seals can remain attractive where space is tight, wiping action matters and shaft speed is moderate. For a fast rotating shaft, however, a labyrinth can reduce solid contact while making the remaining leakage route visible to instruments. For distributed review, an engineering discussion channel can help compare rig observations while controlled data remain in the project’s official repository. The preferred architecture then follows from system requirements rather than a blanket preference for one seal type. It must also return intended oil recirculation to the sump rather than simply minimising every outward flow.
Geometry that controls pressure and leakage
A radial labyrinth begins with teeth cut around the shaft and housing, whereas an axial design places restrictions across a face gap. Tooth count, pitch, flank angle and tooth height shape both the pressure profile and the gap through which oil can pass. Sharp, square teeth are easy to model but can excite the oil film and create vibration; chamfered or tapered teeth trade abrupt pressure changes for smoother flow. The profile should therefore be selected with the lubrication regime in mind.
At high speed, the rotating shaft drags oil tangentially and can establish a low-pressure zone near the teeth. A conventional labyrinth may exploit that effect, whereas a poorly directed spiral can pump lubricant toward the wrong side of the barrier. Engineers should calculate the pressure ratio across each stage and check whether the flow regime supports leakage proportional to viscosity and clearance. Narrow gaps are not automatically superior: their sensitivity to burrs, runout and thermal expansion can outweigh the nominal gain.
Manufacturing method changes the practical meaning of those numbers. Broaching and hobbing suit repeated circumferential teeth, while wire EDM can create fine axial features in harder materials. Any burr, waviness or coating thickness must enter the clearance model rather than being treated as an inspection detail. Current activity described in the recent project update provides context for the wider programme, although individual tooth dimensions must be derived from the target gearbox rather than copied from another machine.
Materials, tolerances and Australian operating conditions
Material selection must balance hardness, toughness, corrosion resistance and dimensional stability. A hard steel or nickel-bearing alloy may suit the shaft, while the housing can use a compatible alloy selected for fatigue strength and ease of manufacture. Surface finishes such as ground or honed tooth flanks can reduce disturbance, but roughness measurements must distinguish the running direction from machining marks. Coatings may improve wear or friction, yet their thickness and build-up need to be included in the running clearance.
Australian operating conditions widen the thermal envelope that the seal must survive. Summer heat around Perth and the Pilbara can raise sump temperature before takeoff, while overnight cooling in regional airfields can expose a parked aircraft to condensation. Coastal salt around Sydney and Melbourne can attack exposed steel if protective finishes are damaged. FIFO maintenance practices also mean a component may move between humid workshops, hot hangars and dusty field environments before returning to service. These realities reward stable clearances over a narrow laboratory optimum.
Manufacturing variation should be represented probabilistically. Monte Carlo analysis can combine tooth pitch, shaft eccentricity, housing bore position, thermal growth and coating thickness into distributions of minimum clearance. Runs should include nominal geometry and credible combinations near the limits, not just a perfectly centred shaft. Probability can be explained with familiar examples before an engineering audience meets Monte Carlo results, although a low-stakes learning resource should be treated as optional context rather than technical evidence. The resulting risk ranking identifies which dimensions need tighter control and whether local compensation or a revised tooth count offers the better robust solution.
Regulatory and environmental duties shape the validation record as well as the hardware. If a seal enters an Australian-registered aircraft, the change must sit within approved design and maintenance instructions under the applicable Civil Aviation Safety Regulations, including Parts 21 and 91 for design approval and maintenance controls. A research rig does not create that approval. Lubricant handling also follows safety data sheets and GHS practice, while waste oil must meet the applicable state or territory environmental requirements. Those records should travel with the test article.
Simulation and tests that earn confidence
Simulation begins with the operating envelope: shaft speed, oil pressure and temperature, housing pressure, radial runout and the intended direction of leakage. CFD can estimate windage, pressure recovery and fluid pumping through the tooth passages. A thermoviscous lubrication model can examine shear in clearances, while finite element analysis captures tooth compliance and thermal distortion. Because each model has assumptions, its predictions need calibration against measured leakage, torque and temperature rise.
Design of experiments should then distinguish robust choices from attractive but sensitive ones. Tooth count, clearance, pressure ratio, speed, oil viscosity and runout can be screened in a fractional design before the most influential factors receive a fuller response-surface study. Replicates are essential because oil temperature and bearing-housing pressure can drift during a run. The measured responses should include external leakage rate, pressure in the protected space, windage torque, temperature rise, vibration and any tooth-contact evidence.
A high-speed test article needs more than a reservoir and a stopwatch. Sample lines should capture oil leaving the seal while vented bearing-housing pressure is measured separately. Shaft speed, torque, oil temperature, vibration and surface temperature should be time-synchronised, with uncertainty derived from calibration records and repeatability tests. For collaborators needing controlled working files, the project member area can provide access to shared protocols and revision history. After testing, the article should be inspected, sectioned where appropriate and compared with the simulated damage modes.
Build, validate and release the seal
Prototype assembly should treat the labyrinth as a system, not a loose pair of toothed rings. Material certificates, machining records, surface finish, coating state and assembly torque belong in the build record. Gauges and fixtures must account for shaft eccentricity and housing alignment, because an incorrectly centred assembly can conceal a design that would perform well in service. First tests should compare the as-built seal with a reference instrumented housing.
A practical release sequence should combine analytical margins with measured evidence:
- Define a leakage budget tied to oil top-up intervals, bearing-housing pressure limits and environmental exposure.
- Model the full tolerance stack, including runout, thermal growth, coating thickness, burrs and housing alignment.
- Test nominal and limit builds across minimum, nominal and maximum shaft speeds.
- Measure leakage, vent pressure, torque, temperature and vibration with calibrated, time-synchronised instruments.
- Compare new hardware with aged, contaminated and reconditioned components representative of service.
- Publish acceptance limits, uncertainty bands, failed samples and the reasons for every design change.
Validation should include an accelerated duty cycle that reproduces thermal soak, pressure pulses and high-speed running. Used oil should be analysed for viscosity, water content and wear debris before and after each campaign. Correlation between simulation and test is more valuable than a single impressive leakage number: it shows which assumptions need revision and where the design has little margin. Failed seals should undergo controlled teardown so that scoring, cracking or debris can be linked to the governing tolerance.
Place a candidate geometry, operating envelope and measurement uncertainty into the OPTIMIZE Project’s design workflow. Review the project videos, objectives and methodology, then bring representative hardware and oil data to the test plan. Share the same definitions of leakage, temperature and failure so results can be compared across simulation and rigs, building a practical case for lower leakage, less heat and dependable service intervals.