How jet count drives oil shear losses in geared engines
Aerospace propulsion is being pushed harder than ever, with designers chasing more power from lighter drivetrains. In Australia's growing aerospace engineering scene, from the workshops of RMIT in Melbourne to the simulation clusters at the University of Sydney, the conversation around geared turbofan architectures has moved well beyond lecture theatres. It is shaping how the next generation of regional aircraft and unmanned platforms will perform on the apron and in cruise. At the heart of that conversation sits a small but stubborn inefficiency: the energy that the oil itself absorbs as it is flung around inside a gearbox.
Oil shear losses, sometimes called churning or windage losses, grow quickly as shaft speeds climb past 20,000 rpm. They can quietly eat several kilowatts of power that the engine would otherwise deliver to the fan or propeller. Designers often respond by adding more lubrication, but more oil is not always better. Each extra jet adds momentum to the sump, stirs the air-oil mixture and changes the thermal balance of bearings and gears. The trade between feed rate and parasitic drag is rarely intuitive, which is why a structured look at jet count is worth doing. A recent design-of-experiments campaign inside the OPTIMIZE project set out to do exactly that, mapping jet count against oil flow rate, shaft speed and lubricant temperature, and watching how the predicted shear losses responded.
That tension between lubrication and loss is what the campaign tried to quantify. Rather than guess the right number of jets, the team built a matrix of cases, ran them through coupled CFD and system simulations, and processed the results with a proper statistical lens. The output is a repeatable response map that engineers can lean on when the trade space feels fuzzy. It also feeds directly into the wider OPTIMIZE drive to cut power loss in the gearbox so the rest of the propulsion system can do its job.
Why jet count is not a free choice
Adding a jet sounds harmless, but each nozzle injects a small parcel of oil at high velocity directly into the mesh region or the bearing chamber. That oil has to be accelerated by the gears and bearings, which means the lubricant itself becomes a load. The shear stress at the gear tooth root and the bearing race is what removes energy from the shaft and turns it into heat. With two jets instead of one, the local oil mass flow doubles in a region already crowded with rotating surfaces.
In a typical high-speed planetary stage, shear losses scale non-linearly with oil flow. Double the flow and you can triple the drag torque once the gears are partly submerged in their own splash. The relationship depends heavily on jet orientation, where the sprays hit the sump, and whether the scavenge ports can keep up. Jet placement and scavenge efficiency are part of the same conversation as jet count, even when the experiment is narrowed to a single variable.
The Australian operating context matters here too. Aircraft on routes between Sydney and Perth spend long stretches at altitude where lubricant temperatures stay low and viscosity stays high. Cold oil shears harder, so a small loss on a warm test cell in Adelaide can become a much bigger parasitic load at cruise. That is why temperature should be swept as a factor in any DOE on this problem.
Building the experiment matrix
A design-of-experiments approach starts with the factors and the levels. The team fixed the gear geometry, lubricant grade and housing layout, then varied jet count across four levels: two, three, four and six jets. Oil flow rate was swept across five settings from a low feed that just kept the bearings wet to a high feed near full immersion. Shaft speed ran at three points covering cruise and take-off, and oil inlet temperature was varied to mimic cold-soak starts versus steady-state cruise.
A full factorial on those factors would have produced hundreds of runs, far too many for the high-fidelity CFD models the team wanted to use. Instead, they used a space-filling Latin hypercube design that captured the corners and middle of the design space without wasting simulations on redundant combinations. Each run was replicated twice to separate numerical noise from genuine physical response, and a few centre points were repeated to track solver drift.
The choice of design mattered because shear loss is driven by interactions. Jet count alone does not predict shear; jet count interacts with flow rate and speed in ways that only show up when the cases are laid out properly. A poorly structured matrix would have hidden those interactions behind apparent main effects and left the team guessing which lever to pull.
Simulating the splash zone
Each case in the matrix was run through a coupled CFD and gear-dynamics model. The CFD domain captured the gear meshing region, the bearing chambers, the jet impingement zones and the scavenge flow back to the tank. A sliding-mesh approach handled the rotating gears, while a free-surface model tracked the oil level in the sump so partially submerged conditions could be resolved.
Shear stress was integrated across the gear and bearing surfaces to give a torque loss, and that loss was added to the gear mesh and bearing drag computed by the system model. The full drivetrain torque budget then showed how much input power was being spent on moving oil rather than turning the shaft. Heat generation was also tracked, because the temperature rise from shear loss feeds back into viscosity, which in turn changes the loss. That coupling was closed iteratively until the case converged.
The rig runs that followed were done on a back-to-back test bed, which let the team measure mechanical loss directly by subtracting electrical input from electrical output. In Australia, facilities like DSTG's Fishermans Bend site and the National Aerospace Testing Alliance at Tullamarine have the high-speed power absorption hardware needed to close a loop like this. The comparison between predicted and measured shear torque gave a confidence interval engineers can carry forward into full-engine integration.
Reading the response surface
Once the matrix was complete, the team fitted a response surface to the data and pulled out the main effects and the interactions. Jet count came out as a strong main effect, with shear loss rising as more jets were added, but the effect was far from linear. Going from two to three jets cost roughly a third as much as going from three to six, which suggests diminishing penalties once enough oil is reaching the critical surfaces.
Oil flow rate interacted strongly with jet count. At low flow, adding a third jet barely changed the loss because there was little oil in the sump. At high flow, the same third jet pushed the loss up sharply because the sump was near its load capacity and the extra jet sprayed oil into regions where the gears were churning air-oil froth. That interaction is the kind of insight a one-factor-at-a-time sweep would miss entirely.
Shaft speed was the dominant driver of absolute loss, but it did not change the ranking of jet count. The best jet count at 18,000 rpm was still the best jet count at 24,000 rpm, just with a bigger loss number attached. That kind of stability is useful because it lets designers pick a jet layout once and trust it across the flight envelope rather than retuning for every mission segment.
Validating on the rig
The simulation campaign was paired with a physical test programme to make sure the trends held up in metal. Two gearbox builds were run on the same back-to-back rig: one with the baseline four-jet configuration and one with the lower-loss three-jet layout identified in the simulation. Oil temperatures were held at three steady points to match the matrix, and shaft speed was swept across the same range as the CFD runs.
Measured shear torque tracked the predicted values within a tight band across most of the operating range. The largest deviations appeared at the lowest speeds and lowest flow rates, where the CFD free-surface model struggled to resolve very thin oil films. Even there, the trend with jet count was preserved, which is what mattered for the design decision. The team also ran a tolerance study, varying jet hole diameter and spray angle within manufacturing limits, and found that the optimal jet count was robust to those small geometric perturbations.
A quick look at how this lines up with regional realities helps. Operators running turboprops out of places like Wagga Wagga or Cairns care about fuel burn per sector, and every watt of parasitic loss in the gearbox shows up directly on the fuel bill. A rig-validated result that says three jets are enough for the cruise condition gives those operators a lighter oil system, a smaller scavenge pump and a quieter gearbox without giving up bearing life. That combination is hard to argue with when the books are being balanced at the end of the quarter, and it fits neatly alongside the wider OPTIMIZE work on drivetrain losses, including a companion piece on elastomeric couplings that handles related torsional issues in the same drivetrain.
What this means for designers
The practical takeaway from the campaign is a simple rule of thumb wrapped in a clear caveat. For a planetary stage running in the 20,000 to 25,000 rpm window, three well-placed jets gave the best balance of lubrication and shear loss in the tested geometry. Four jets added measurable loss without a corresponding gain in bearing or gear temperature. Six jets pushed the sump into a churning regime that hurt efficiency without helping reliability.
The caveat is geometry. A different gear module, housing shape or oil grade will shift the sweet spot, which is why the team recommends rerunning the matrix whenever any of those parameters change. The methodology scales; the absolute answer does not.
For engineers working on next-generation power reduction gearboxes, the study is a reminder that structured experimentation pays off where intuition struggles. Churning losses are sensitive, multi-factor problems, and a well-designed DOE turns them into something that can be reasoned about rather than guessed at. The wider OPTIMIZE work on tolerance analysis, hyperstatic load paths and power density all benefit when the lubrication system is no longer the unknown element of the gearbox.
Key experimental factors swept in the study
- Jet count: two, three, four and six lubrication jets
- Oil flow rate: five settings from minimal feed to near-immersion
- Shaft speed: three points covering cruise and take-off
- Oil inlet temperature: cold-soak, mid and steady-state cruise values
- Jet hole diameter and spray angle, varied within manufacturing tolerance
Practical observations from the response surface and rig runs
- Three jets gave the best loss-versus-lubrication trade in the tested geometry
- Adding a fourth jet raised shear loss without reducing bearing or gear temperature
- The optimal jet count was stable across the simulated shaft-speed range
- Tolerance variation in jet diameter and angle did not change the ranking
- Predicted and measured shear torque agreed within a tight band for most operating points
Engineers who want to dig into the underlying data and supporting rig logs can sign in through the members area of the OPTIMIZE project portal, where the full report and simulation files are available. Pairing this study with the rest of the OPTIMIZE toolkit gives design teams a clear path to lower loss and lighter gearboxes, with measurable gains in fuel burn on every sector from a Sydney morning departure to a Perth late afternoon arrival.