Direct Oil Cooling in Axial Flux Motors: Dielectric Fluid Compatibility, Flooded Air-Gap Churning Losses, and Stator Potting Sourcing

Technical review: 31 October 2026. Scope: English-language guidance for global OEM procurement and engineering teams comparing liquid-cooled traction and industrial AFPM designs. Aerospace applications require additional mission and certification work. This is a design-screening and sourcing guide, with derived calculations and illustrative assumptions, not measured AFPM Motor product data. It does not establish fluid approval, insulation lifetime, or a motor rating. Review again by April 2027, or when fluid grade, inverter waveform, winding materials, or maximum speed changes.

For an architecture review, share your duty cycle and coolant limits with our AFPM Motor OEM Manufacturing team and agree the evidence needed through AFPM Quality Validation.

1. The Thermal Bottleneck in Sandwich AFPM Stators

In a dual-rotor, single-stator AFPM machine, a central winding assembly sits between rotating magnet discs. A perimeter jacket may leave a long conduction path from inner winding regions to the housing. Slot insulation, contact interfaces, potting thickness, and mounting paths can govern the hotspot even when coolant outlet temperature is low.

Example perimeter-jacket heat path:
Copper hotspot → wire insulation → core or potting → housing rim → coolant

Example direct-oil heat path:
Copper hotspot → wire insulation / local potting → wetted surface → oil

Direct cooling reduces selected thermal resistances; it does not remove wire insulation or guarantee that every turn is wetted. A sealed stator circuit can also cool the winding without flooding the rotor gap. The IEEE proof-of-concept study demonstrates liquid cooling in an axial-flux machine using coolant conduits integrated with tooth coils and PAO oil. That architecture is evidence of feasibility, not a benchmark for spray or flooded-gap performance.

Ask suppliers to compare winding hotspots at the same torque, speed, coolant inlet temperature, flow, and ambient conditions. Continuous current density cannot be inferred from cooling topology alone. At fixed conductor area and resistance, doubling current quadruples copper loss; resistance itself rises with winding temperature. A thermal model and stabilized duty-cycle test are needed to translate cooling into a continuous rating.

2. Architecture Comparison: Flooded Cavity vs. Scavenged Spray vs. Canned Stator Sleeve

Cooling topology sets the hydraulic, magnetic, and sealing interfaces. The diagram shows conceptual configurations; the matrix records engineering trade-offs rather than catalogue performance promises.

Detailed Trade-Off Matrix

These eight criteria are a comparison framework. Heat-transfer coefficients, current density, drag, and speed limits require geometry-specific evidence.

3. The Physics of Air-Gap Churning & Viscous Windage Losses

A rotating disc shears fluid against the stationary stator. Oil accumulation can add drag and heat, but the amount depends on viscosity, temperature, rotor dimensions, clearances, fill fraction, and surface features. Axial rotor-stator cavity flow should not be treated as a universal cylindrical Taylor-Couette correlation.

Drag Torque and Power Scaling Formulations

For an ideal, fully wetted annular face with linear laminar shear, no slip, and uniform axial gap, integrating local shear gives:

M_face = π μ ω (Ro⁴ − Ri⁴) / (2 g)
P_faces = N M_face ω
Re = ρ ω Ro² / μ

μ: dynamic viscosity [Pa·s]; ρ: density [kg/m³]
ω: angular speed [rad/s] = 2π × RPM / 60
g: axial gap [m]; Ro, Ri: outer and inner wetted radii [m]
N: number of wetted rotor faces opposing a stationary surface

A dual-rotor, single-stator layout has two inner rotor faces opposing the stator. Outer faces and rims need separate treatment. This simplified model gives power proportional to viscosity, speed squared, and the fourth-power radius difference. It omits pumping, free surfaces, centrifugal circulation, and turbulent effects, so it is not a validated high-speed drag prediction.

For other regimes, a dimensionless moment coefficient can be defined through P = 0.5 × CM × ρ × ω³ × Ro⁵ per face. The coefficient depends on Reynolds number, gap ratio, enclosure, and flow conditions; holding it constant is an assumption, not a universal cube-of-speed or fifth-power-radius law. Do not approve a high-speed design using an unspecified coefficient or a fixed Reynolds transition threshold.

A Reproducible Drag-Budget Example

If a supplier test reports added drag torque, the power conversion is exact: added loss = added torque × angular speed. The table assumes a constant 0.50 Nm added torque solely to illustrate the conversion; it is not oil, air, or motor test data. Real drag torque changes with speed and fluid condition.

These percentages express a loss budget relative to output, not percentage-point efficiency changes. For example, a project allowing 1.5 kW added drag at 8,000 RPM permits approximately 1.79 Nm of added torque. That is a buyer-selected budget, not an industry acceptance threshold.

Sourcing Takeaway on Fluid Dynamics

Request no-load torque-versus-speed measurements across the specified oil-temperature range, with matched bearing and seal conditions. Record fill level, fluid grade, flow, stabilization time, instrument uncertainty, and repeatability. A flooded-versus-drained difference can include changes in bearing lubrication and sealing; document those confounders before calling it isolated air-gap churning.

Compare winding-temperature improvement against added rotor drag, feed/scavenge pump input, and heat-exchanger demand over the duty cycle. Select scavenged spray or sealed stator cooling when measured system results justify it; there is no universal speed above which flooding is forbidden.

4. Dielectric Fluid Selection & Material Compatibility Matrix

Specify a fluid product and revision, not just an ATF, PAO, or EV-fluid family. Base chemistry alone does not establish electrical performance, additive compatibility, viscosity, or seal life. A shared gearbox sump also introduces gear-wear debris and lubricant requirements.

Three Primary Classes of Dielectric Fluids

1. Transmission / shared-sump fluids: May satisfy gearbox lubrication requirements, but need motor-specific copper, insulation, seal, and electrical-property evidence. An ATF label neither approves nor disqualifies a fluid for an AFPM motor.
2. PAO-based formulations: Candidates for cooling circuits; final additives and material interactions matter. Low moisture affinity does not make a circuit immune to water contamination or electrical breakdown.
3. Dedicated e-drive cooling fluids: Request the exact grade's property sheet and application approval. Do not infer a guaranteed drag reduction, copper rating, or inverter compatibility from EV branding.

Fluid Property and Compatibility Evidence

Use this six-row comparison when requesting data for any of the three fluid families. Compare like-for-like methods and conditions rather than unsourced family-average values.

For liquid breakdown measurements, ASTM D1816 covers insulating liquids. A laboratory result is not a motor voltage rating or proof of cleanliness. Specify the method's applicability with the fluid supplier and insulation engineer, including aged and contaminated fluid conditions.

Stator Material Chemical Compatibility

ASTM D3455-11(2019) addresses screening with petroleum-origin insulating oil. Its scope does not automatically qualify every synthetic e-drive fluid or complete insulation system, and a screening pass does not establish equipment suitability or lifetime. Agree suitable exposures and acceptance criteria for the actual formulation.

1. Magnet wire enamel: Record the wire supplier, insulation build, thermal class, and inverter exposure. PAI-based and PEEK systems are candidates, not guarantees of oil immunity; qualify aged coupons and representative windings.
2. Slot liners, tapes, seals, and magnet adhesives: Include every wetted adhesive and coating in testing. Measure swelling, hardness, bond retention, and electrical properties as appropriate; a generic material name does not approve its adhesive system.
3. Stator potting resins: Request cured thermal conductivity, cure schedule, glass-transition behavior, CTE, bond strength, and changes after fluid exposure. Filled epoxies can improve conduction, but viscosity, voids, thermal stress, and repairability also affect selection. Do not require exact CTE matching or a universal minimum conductivity without a thermal and structural model.

For converter-fed Type I insulation, IEC 60034-18-41 defines qualification and quality-control testing. Confirm that Type I is appropriate for the design; the reference does not certify a fluid or guarantee partial-discharge-free operation for every inverter waveform.

Potting sourcing decision: Reject a resin quotation that provides only a dry, uncured, or unspecified-temperature conductivity value. Ask for production-cure data, fluid-aged adhesion and insulation results, material traceability, and agreed void-inspection criteria on a representative stator. Compare complete thermal paths, not resin conductivity alone.

Before freezing the bill of materials, use AFPM Quality Validation to discuss which coupon, winding, and complete-motor evidence the program needs.

5. Mechanical & Hydraulic Integration: Seals, Scavenging, and Filtration

Dynamic Shaft Seals at High Peripheral Speeds

Seal surface speed is v = π × shaft diameter × RPM / 60. A 45 mm shaft at 9,000 RPM has a surface speed of approximately 21.2 m/s. Check that point against the actual seal's pressure, temperature, fluid, shaft-finish, and runout envelope. Neither an elastomer family nor a PTFE label supplies a universal speed rating. Labyrinth or slinger arrangements also need demonstrated containment and contamination control.

Scavenge Pump Sizing: Account for Entrained Air

If the return mixture has liquid volume fraction α, its approximate volumetric flow at the pump inlet is Qmixture = Qliquid / α. For example, 10 L/min liquid at 40% liquid volume fraction requires 25 L/min mixture capacity before margin. This explains a possible 2.5:1 capacity ratio; it is an assumed operating point, not a rule for every AFPM motor.

Specify inlet pressure, air-handling performance, pickup geometry, venting, startup behavior, and all required attitudes and accelerations. Verify that oil does not accumulate against a rotor while avoiding pump cavitation and excessive air ingestion. Choose sump pressure and capacity from the tested circuit rather than prescribing a universal vacuum target.

Oil Cleanliness & Air-Gap Particulate Filtering

Bearing and gear debris can collect near magnets or damage insulation and clearances. Agree cleanliness limits, filter efficiency at a stated particle size, pressure drop, cold-start bypass behavior, and inspection intervals. A micron label alone is insufficient. A magnetic trap can capture ferrous debris, but does not replace filtration or detect water and non-ferrous contamination.

6. Sourcing & RFQ Engineering Checklist

Define pass/fail limits jointly before ordering prototypes. The checklist identifies evidence to obtain; values must follow the project's electrical, thermal, and mechanical envelope.

• 1. Cooling architecture: Identify flooded, scavenged spray, or sealed stator cooling. Supply sections, wetted areas, fill-level limits, orientation, and duty cycle.
• 2. Drag and system-loss budget: Obtain hot/cold spin-loss maps, measurement uncertainty, test boundary conditions, and feed/scavenge pump power. Compare against an agreed duty-cycle loss budget.
• 3. Exact fluid specification: List approved part number and revision, viscosity-temperature data, electrical test conditions, handling requirements, and substitution approval process.
• 4. Material compatibility: Obtain a wetted-material bill of materials and an agreed aging protocol covering fluid and material changes. Record exposure time, temperature, acceptance limits, and scope of each cited standard.
• 5. Potting process: Confirm production cure, cured conductivity, glass-transition behavior, thermal expansion, aged adhesion, inspection method, and lot traceability. Distinguish encapsulation from impregnation processes.
• 6. Hydraulic interfaces: Specify ports, pressure-drop map, supply flow, return air fraction, pump maps, and drainage verification at required pitch, roll, and acceleration.
• 7. Shaft seals and rotor materials: Record seal part numbers, speed/pressure/temperature ratings, shaft tolerances, magnet coating, adhesive compatibility, and endurance evidence.
• 8. Acceptance tests: Agree thermal stabilization, torque-speed points, leak-test pressure and allowable leakage, insulation withstand and partial-discharge tests where applicable. Separate prototype qualification from production routine tests.

7. Frequently Asked Questions (FAQ)

Can we share the oil loop between the axial flux motor and an integrated reduction gearbox?

Potentially, if the same formulation meets gear lubrication and motor-material requirements. Assess debris transport, filtration, air release, and electrical properties after aging. Separate circuits can reduce cross-contamination but add pumps, mass, and service interfaces; they do not guarantee dielectric performance.

What flow rate would a 150 kW output AFPM motor need?

For 150 kW shaft output at 96% motor efficiency, input power is 156.25 kW and motor losses are 6.25 kW, excluding external pump and inverter losses:

Ploss = Pout × (1 / efficiency − 1)
      = 150 kW × (1 / 0.96 − 1) = 6.25 kW

Assumptions: 75% of motor loss enters the stator oil branch;
oil density ρ = 830 kg/m³; heat capacity cp = 2,100 J/(kg·K);
inlet-to-outlet rise ΔT = 15 K.

Qstator = 4,687.5 W / (830 × 2,100 × 15)
        = 0.0001793 m³/s ≈ 10.8 L/min

If this same loop rejects all 6.25 kW: Qtotal ≈ 14.3 L/min.

These are illustrative bulk heat-balance flows, not pump specifications or guaranteed hotspot performance. Obtain fluid properties at operating temperature, establish where rotor and other losses go, and check nozzle distribution, pressure drop, bypass flow, and transient heat storage.

How much auxiliary power does a scavenge pump consume?

At 25 L/min mixture flow and 50 kPa pressure rise, the simple incompressible estimate Phydraulic = Δp × Q is 20.8 W. At an assumed 35% total efficiency, electrical input is approximately 59.5 W. Aerated flow, suction conditions, and gas compression require a measured pump map; this example is not a pump performance claim. Compare feed and scavenge pump input with the measured drag reduction over the duty cycle before claiming a net saving.

Does dielectric oil in the air gap change torque or inductance?

For a nonmagnetic fluid replacing air without changing geometry, static magnetic reluctance is approximately unchanged. That does not guarantee identical operating torque or inductance: temperature, saturation, mechanical clearances, and any containment wall matter. Oil and air mixtures also change electrical permittivity, so assess capacitance, common-mode currents, and inverter insulation stress with the actual geometry and fluid.

Can direct oil cooling prevent thermal demagnetization of NdFeB rotor magnets?

Direct wetting can improve rotor heat removal when the design delivers coolant to the relevant surfaces. It does not guarantee a magnet temperature or permit a lower magnet grade without validation. Check magnet loss, local temperature, demagnetizing field, grade-specific coercivity, coating, and adhesive aging. A sealed stator circuit may provide little direct rotor cooling.

What happens if the dielectric oil degrades?

Oxidation, contamination, or additive changes can alter acidity, deposits, viscosity, dielectric behavior, and material compatibility. Trend samples against supplier-approved limits for the exact fluid, with attention to water and wear debris. Set sampling, fluid replacement, alarms, and thermal derating from endurance evidence and the duty cycle; there is no universal oxidation temperature or service interval for all e-drive oils.

8. Sourcing & Engineering Summary

1. Compare total system performance: Require winding hotspots, rotor temperatures, drag, and pump input over the specified duty cycle.
2. Approve a fluid-material combination: Use exact product identities and controlled test conditions; neither fluid-family names nor standalone breakdown kV values establish approval.
3. Source potting as a qualified process: Obtain production-cure and fluid-aged evidence on representative stators, including interfaces and voids.
4. Freeze measurable acceptance criteria: Agree hydraulic, sealing, thermal, and electrical tests before comparing RFQ prices or releasing a pilot order.

Sources & References

Source review: 31 October 2026. The references below support cooling feasibility and test-method scope. They do not establish universal AFPM current density, drag, fluid-property, or lifetime benchmarks. Public abstracts and catalogue scope statements were used; full standards and project-specific testing are needed for qualification.

1. IEEE Transactions on Energy Conversion (2016). Direct Liquid Cooling in Low-Power Electrical Machines: Proof-of-Concept, DOI: 10.1109/TEC.2016.2597059. Uses an axial-flux prototype with coolant conduits and PAO oil; supports feasibility rather than flooded-gap or spray performance claims.
2. ASTM International. D3455-11(2019): Compatibility of Construction Material with Electrical Insulating Oil of Petroleum Origin. Material/oil screening scope; a screening result alone does not prove equipment suitability or lifetime.
3. ASTM International. D1816: Dielectric Breakdown Voltage of Insulating Liquids Using VDE Electrodes. Liquid dielectric test method; document electrode spacing and sample conditions before comparing results.
4. International Electrotechnical Commission. IEC 60034-18-41:2014: Converter-fed Type I winding insulation qualification and quality-control tests. Relevant to inverter-fed insulation-system evaluation; does not approve oil chemistry or replace machine-specific validation.

Need a Cooling Architecture Review for Your AFPM Program?

Send your continuous and peak torque-speed duty, DC bus and inverter waveform, coolant inlet limits, candidate fluid, and packaging constraints to our engineering team. We can discuss the manufacturing scope and validation evidence needed for an RFQ.

• Review AFPM Motor OEM Manufacturing for prototype and production planning.
• Use AFPM Quality Validation to scope acceptance records and supplier test coordination.
• Discuss magnet coating, bonding, and retention interfaces through Rotor & Magnet Assembly.