How a magnetic coupling works
Two magnet assemblies, one on the driving shaft and one on the driven shaft, separated by a non-magnetic containment shell. The magnetic circuit crosses the shell; the fluid does not.
| Type | Principle | Slip | Where used |
|---|---|---|---|
| Synchronous | Permanent magnets on both rotors lock pole to pole | None below breakaway | Sealless pumps, agitators, compressors — the standard for power transmission |
| Eddy current / hysteresis | Magnets on one side induce currents in a conductive rotor | Inherent, speed-dependent | Variable-speed drives, soft starting, torque limiting |
| Axial (face) | Discs facing across an axial gap | None | Short axial envelopes, simple assemblies |
| Coaxial (radial) | Concentric cylinders | None | Higher torque density; the pump standard |
Why the seal removal matters commercially
Mechanical seals are the leading cause of centrifugal pump failure, and on hazardous, toxic or valuable fluids they are also the leading source of fugitive emissions and product loss. A sealless magnetic drive removes the failure mode entirely rather than managing it: no seal faces to wear, no flush plan to maintain, no seal support system, and no leak path to atmosphere. On duties where a seal failure means an environmental release or a plant shutdown, that changes the total cost picture regardless of the coupling’s higher purchase price.
The containment shell and eddy losses
The shell is where magnetic coupling design gets genuinely difficult, and where most first-time designs go wrong.
A metallic shell rotating in a magnetic field is a short-circuited transformer
The outer magnet set sweeps an alternating field across the shell. In any electrically conductive shell that induces circulating eddy currents, which do three things: they consume power that never reaches the impeller, they generate heat inside the containment where cooling is poor, and they produce a drag torque that shows up as reduced efficiency.
The loss scales with the square of the field, with the square of the frequency — so with pole count and speed — and with shell thickness and conductivity. On a high-speed, high-pole-count coupling it can become a significant fraction of transmitted power, and the resulting heat has to be carried away by process fluid that may not be flowing much.
| Shell material | Eddy losses | Trade-off |
|---|---|---|
| Austenitic stainless (316, Hastelloy) | Significant | Strong, corrosion resistant, weldable, familiar to fabricators and inspectors. The default where losses are tolerable |
| Titanium | Lower | Higher resistivity than stainless so fewer losses; excellent corrosion resistance; more expensive and harder to fabricate |
| Ceramic | Essentially zero | Non-conductive, so no eddy losses at all; brittle, and pressure containment in a brittle material needs careful design |
| Fibre-reinforced composite | Essentially zero | Non-conductive and light; temperature and chemical compatibility limits, and pressure rating is design-specific |
Design consequences
- Shell thickness is a direct torque tax. It sits in the magnetic gap, and field falls steeply with distance. Pressure rating drives thickness up; magnetic performance drives it down. That tension is the central design trade.
- Pole count is not free. More pole pairs raise torque density for a given diameter, but they also raise the switching frequency the shell sees, and eddy losses scale with the square of it. There is an optimum, and it is application specific.
- Consider the loss at the design stage, not after testing. A coupling that meets its torque requirement on a bench with a non-conductive shell and then loses several percent to a stainless production shell is a common and avoidable surprise.
- Heat has to go somewhere. Verify the thermal path for eddy losses under the worst realistic case — which is usually low flow, not full flow.
Torque, breakaway and decoupling
A synchronous coupling transmits torque with zero slip up to a limit — and then loses it completely. There is no partial engagement and no graceful degradation.
Size against the real worst case
Selecting a coupling on normal running torque is the classic error. Size against whichever of these is largest, with margin:
- Starting torque, particularly a flooded start or a viscous fluid at low temperature
- Viscosity at the coldest operating condition, not at process temperature
- Any transient — a valve slam, a partial blockage, solids ingress
- Torque derating at maximum operating temperature, since transmitted torque falls with magnet remanence
And then confirm the protection strategy: many installations add power monitoring or temperature sensing on the shell precisely because decoupling is otherwise silent until something fails.
Temperature and torque
Transmitted torque tracks magnet remanence, and remanence falls with temperature. A coupling sized at ambient will transmit measurably less at process temperature — roughly 12% less for NdFeB across a 100 °C rise, reversibly. Two rules follow:
- Size the torque at the maximum operating temperature, not at 20 °C.
- Check the temperature class against irreversible loss. If the magnets exceed their rated maximum, torque capacity falls permanently and the coupling will decouple at a load it used to carry. That failure looks like a process problem and is actually a magnet problem. The load-line method is in the temperature guide.
- Above roughly 150 °C, samarium cobalt becomes the sensible choice — both for headroom and because its far lower temperature coefficient means torque stays predictable across the range.
Where they are used
| Application | Why sealless | Specification driver |
|---|---|---|
| Chemical process pumps | Toxic, corrosive or valuable fluid with no acceptable leak path | Chemical compatibility of the shell; torque at process temperature |
| Pharmaceutical and biotech | Sterility and contamination control; no seal flush into product | Cleanability, material certification, validation documentation |
| Refrigeration and HVAC compressors | Hermetic refrigerant containment, no seal leakage of working fluid | Refrigerant and oil compatibility; efficiency |
| Chilled water and heat transfer | Maintenance elimination on inaccessible plant | Efficiency over a long duty life |
| Food and beverage | No seal leakage into product; cleanability | Sanitary construction — see food processing |
| Semiconductor and ultrapure | Zero particulate generation from a wearing seal | Material purity and cleanliness |
| Marine and subsea | Torque through a pressure hull without a shaft penetration | Corrosion; pressure rating of the shell |
| Agitators and mixers | Sealed vessel, top or bottom entry | Starting torque into a settled or viscous batch |
The adjacent case: sealless motors
A canned motor pump takes the same idea further — the motor rotor itself runs inside the process fluid behind a can, and there is no coupling at all. The magnetic design problem is similar, the eddy loss question is the same, and the fluid compatibility requirement on the rotor materials becomes considerably more demanding. Where a magnetic coupling isolates the fluid from the driver, a canned motor isolates it from the stator only.
Specifying the magnet set
What matters in a coupling magnet set that does not matter elsewhere
- Consistency across the set is as important as absolute strength. A coupling is a matched pair of pole arrays. Variation between magnets within a set produces torque ripple, vibration and uneven loading — specify a total moment tolerance across the set, not merely a minimum per magnet.
- Pole alignment and placement. Angular placement error in either rotor reduces peak transmitted torque and introduces ripple. Positional tolerance in the carrier is as important as the magnet tolerance.
- Retention against a decoupling event. Retention should be designed for the abnormal case — the rapid heating and reversing forces of a decoupling — not just for normal running. Geometry should hold the magnets if adhesive fails. See assembly and retention design.
- The inner rotor may be fluid-wetted. Where it is, coating selection is a chemical compatibility question, not a general corrosion question — and encapsulation of the whole inner assembly is common rather than relying on per-magnet plating.
- Stray field outside the coupling. The outer rotor is not fully enclosed in every design, and the field around it can affect nearby instruments and sensors. Containing it is a flux-circuit problem — see shielding and flux containment.
What we supply
- Arc segments, blocks and rings for inner and outer rotor sets, in matched batches.
- NdFeB in H through EH classes for standard and elevated-temperature service.
- Samarium cobalt for high-temperature couplings and where intrinsic corrosion resistance is an advantage.
- Coating and encapsulation specified against the process fluid rather than against a generic corrosion class.
- Total moment matching across a set with helmholtz measurement, where torque ripple matters.
What to send us
- Required transmitted torque, and the worst case — starting, cold viscosity, transient
- Operating speed and the pole count if already chosen
- The magnetic gap: shell thickness plus both running clearances
- Shell material
- Maximum operating temperature at the magnets
- Process fluid, and whether the inner rotor is wetted
- Envelope: outer rotor OD, inner rotor ID, axial length available