Fluid Coupling

What a fluid coupling actually is

Take two identical bowls, each lined on the inside with straight radial vanes — think of half a doughnut sliced through its equator, paved with blades. Bolt one bowl to your engine's output shaft and the other, facing it, to the shaft you want to drive. Leave a few millimetres of gap between them, seal the pair inside a housing, and fill it with light hydraulic oil. You have just built a fluid coupling. The driving bowl is the impeller (also called the pump); the driven bowl is the turbine (also the runner). Nothing solid bridges the gap. Torque crosses it entirely through the oil.

The mechanism is the same physics that makes a centrifugal pump work. When the impeller spins, each pocket of oil between its vanes is flung radially outward — pressed against the outer rim by centrifugal force. The oil has nowhere to go but across the gap into the turbine, which it enters near the outer rim moving fast and in the direction of impeller rotation. Striking the turbine's vanes, the oil pushes them around, surrenders most of its angular momentum, and is left moving slowly. It then drifts inward along the turbine vanes toward the axis, crosses back into the impeller near the centre, and is re-accelerated. Repeat at thousands of revolutions per minute and you have a continuous toroidal vortex: oil corkscrewing around the ring of blades, simultaneously rotating with the assembly and circulating through it.

That circulating flow is the whole game. The torque delivered to the turbine equals the rate at which the fluid's angular momentum changes as it passes through — the oil leaves the impeller with high tangential momentum and leaves the turbine with low tangential momentum, and the difference is exactly the torque handed off. No flow, no torque. This is why a fluid coupling is sometimes called a hydraulic coupling, a fluid flywheel, or a Föttinger coupling after Hermann Föttinger, the German engineer who patented the architecture in 1905 for marine propulsion.

Slip — the feature, not the bug

Here is the counter-intuitive heart of the device: a fluid coupling only works because the two halves run at different speeds. If the impeller and turbine turned at precisely the same rate, the oil between them would rotate as one rigid block. There would be no radial circulation, no momentum exchange, and therefore no transmitted torque. The speed difference is what drives the vortex.

That speed difference is called slip, defined as a percentage:

slip  s = (N_in − N_out) / N_in × 100 %

At standstill, before the load has begun to move, slip is 100 percent — the impeller spins, the turbine is stationary, and the motor can wind up to nearly full speed against an almost-unloaded shaft. As the turbine accelerates, slip falls. At rated load a well-designed industrial coupling settles at roughly 2 to 4 percent slip. It can never reach zero, because zero slip means zero circulation means zero torque. The coupling lives in a narrow band: enough slip to circulate oil and carry the load, little enough that not too much power is wasted as heat.

The transmitted torque scales with the square of speed and the fifth power of the coupling's working diameter — the same dimensional law as any centrifugal turbomachine:

T = λ · ρ · ω² · D⁵

where  T = torque (N·m)
       λ = dimensionless torque coefficient, a function of slip
       ρ = oil density (≈ 870 kg/m³ for typical hydraulic oil)
       ω = impeller angular velocity (rad/s)
       D = working circuit diameter (m)

The fifth-power dependence on diameter is brutal and useful: a small increase in size buys a large jump in torque capacity, which is why coupling catalogues step diameters in fine increments. The λ(slip) curve rises steeply from zero at full lockup to a peak somewhere near 80–100 percent slip — meaning the coupling produces its highest torque at start-up, which is exactly the wrong way around for a clutch but exactly right for breaking away a heavy stalled load.

Where the lost power goes

Slip is not free. Every watt of slip becomes heat in the oil. The lost power is simply input torque multiplied by the speed difference:

P_loss = T · (ω_in − ω_out) = P_in · s

At 3 percent rated slip the coupling is about 97 percent efficient — a 200 kW conveyor drive sheds roughly 6 kW continuously into the oil, a manageable trickle that an oil-to-air radiator or a finned housing can dissipate. The danger is the transient. During start-up, and catastrophically during a stall, slip rockets toward 100 percent and the entire input power converts to heat. A crusher that jams while the motor keeps turning can pour hundreds of kilowatts into a few litres of oil in seconds. For exactly this reason large couplings carry fusible plugs: a low-melting-point alloy that melts at a set temperature (commonly around 140 °C), blows out, and dumps the oil to disengage the drive and prevent a fire or a burst housing. Continuous-duty and frequent-start applications add deliberate cooling — circulating oil loops, water jackets, or external heat exchangers — and the heat budget, not the torque rating, usually decides the coupling size for stall-prone duties.

Fluid coupling versus torque converter

The fluid coupling has exactly two working members. Add a third — a bladed stator, also called the reactor, mounted on a one-way (overrunning) clutch in the path where oil returns from the turbine to the impeller — and you have a torque converter. The stator catches the oil leaving the turbine and redirects it so that it re-enters the impeller already moving in the direction of rotation, helping to spin it rather than dragging on it. This recycled angular momentum lets the converter deliver more torque to the turbine than the engine feeds the impeller. It multiplies torque, typically by 1.8:1 to 2.5:1 at stall, fading to 1:1 as the turbine catches up and the one-way clutch lets the stator freewheel.

That difference — two members and a flat 1:1 torque ratio versus three members and torque multiplication — is the whole reason both devices exist. The automotive automatic transmission needs the launch torque boost, so it uses a converter. A conveyor or a fan does not need torque multiplication; it needs a gentle start and protection from shock, so it uses the simpler, cheaper, slightly more efficient plain coupling.

Property	Fluid coupling	Torque converter	Friction clutch
Working members	Impeller + turbine	Impeller + turbine + stator	Friction discs, no fluid
Torque multiplication	None (1:1)	≈ 1.8–2.5:1 at stall	None (1:1)
Steady-state slip	2–4 %	2–5 % (more at stall)	0 % (engaged)
Wear surfaces	None (oil only)	None (oil only)	Friction lining, wears out
Soft start	Excellent	Excellent	Operator-dependent
Vibration isolation	Excellent	Excellent	Poor (rigid when engaged)
Peak efficiency	96–98 %	85–92 % (no lockup)	≈ 100 % (engaged)
Typical use	Conveyors, crushers, fans, marine	Automatic transmissions	Manual transmissions

Worked example — soft-starting a 250 kW belt conveyor

Consider a long overland belt conveyor driven by a 250 kW, 4-pole induction motor running at 1,480 rpm full load. Started across the line with a direct coupling, the motor would snatch the belt with five to seven times rated torque, shock-loading the gearbox and risking belt slip or a snapped splice. Insert a constant-fill fluid coupling and the start changes character entirely.

Motor:            250 kW, 1,480 rpm full load
Coupling fill:    ~80 % of working circuit volume
Rated slip:       3 %  →  output ≈ 1,436 rpm at full load

Start sequence:
  t = 0      motor energised, impeller spins up, turbine still
             slip = 100 %, transmitted torque rises gently
  t ≈ 1 s    motor near full speed, belt begins to move
  t ≈ 8–15 s belt fully up to speed, slip falls to ~3 %

Steady-state heat load:
  P_loss = P_in × s = 250 kW × 0.03 = 7.5 kW into the oil
  (dissipated by the finned housing + airflow)

Peak start torque seen by gearbox:
  ~1.5–1.7 × rated  (vs ~5–7 × for a rigid coupling)

The motor reaches running speed before the belt is loaded, drawing its inrush current only briefly and against low torque, so the electrical supply and the motor windings are spared. The belt accelerates over ten seconds or so instead of a fraction of one. A delay-fill or traction coupling refines this further: an internal reservoir holds part of the oil out of the working circuit at start, then meters it in over a few seconds so the load breaks away even more gently and several motors on one conveyor can be brought up in sequence. The 7.5 kW of continuous slip loss is the running tax for never again shock-loading the drive train.

Variants — constant fill, variable fill, delay fill

• Constant-fill (fixed-fill) coupling. The simplest and most common. A sealed oil charge, typically 80 percent of working volume, set once at assembly. Gives a fixed slip-versus-load characteristic. Used wherever a plain soft start and vibration isolation are all that's needed — fans, pumps, conveyors, compressors.
• Delay-fill (traction) coupling. Adds an internal annular reservoir that withholds part of the oil from the working circuit during start, then lets it flow in under centrifugal action over a few seconds. Extends the soft-start ramp and limits start torque on long, heavily loaded conveyors. Common on mine and quarry belts.
• Variable-fill (scoop-trim) coupling. A movable scoop tube draws oil out of, or lets it back into, the working circuit while running, varying the oil quantity in real time. Because torque capacity falls with reduced fill, this gives stepless control of output speed from a fixed-speed motor — a hydraulic variable-speed drive. Used on large boiler-feed pumps and induced-draught fans in power stations, where it competes with the variable-frequency drive.
• Coupling with delay chamber and fusible protection. Heavy-duty mill and crusher couplings combine a delay chamber, fusible plugs, and sometimes a stall-detection thermal switch. Sized by heat capacity for repeated stalls.
• Fluid flywheel. The historical passenger-car form — a constant-fill coupling used as the launch element ahead of a Wilson preselector epicyclic gearbox (Daimler, Armstrong-Siddeley, 1930s–1950s). Let a car idle in gear and pull away with no friction clutch. The direct ancestor of the automatic transmission.

Where fluid couplings actually show up

• Belt conveyors. The signature application. Overland conveyors in mining and bulk handling, sometimes kilometres long, use delay-fill couplings to stage the start of multiple drive motors and protect splices and gearboxes. Without them, starting a loaded kilometre of belt would routinely snap it.
• Crushers, mills and shredders. Jaw crushers, ball mills, and rotary shredders stall regularly when they bite something too hard. The fluid coupling absorbs the shock, lets the motor keep turning, and protects against the torque spike. Fusible plugs guard against a sustained stall.
• Centrifugal pumps and fans. Variable-fill scoop-trim couplings give stepless speed control on power-station boiler-feed pumps (tens of MW) and large draught fans, regulating flow without throttling losses.
• Marine propulsion. Föttinger's original 1905 application. Fluid couplings (and the related fluid drives) isolate the propeller shaft from engine torsional vibration and allow smooth clutching of diesel engines to the shaft.
• Diesel-electric and railway auxiliaries. Cooling-fan drives and generator drives on locomotives use fluid couplings to decouple speed and damp vibration.
• The fluid flywheel transmission. 1930s–1950s British and German automobiles paired a fluid coupling with a preselector gearbox — the practical first step toward the modern automatic.

Failure modes — where fluid couplings actually break

• Overheating from sustained slip. The single biggest hazard. A jammed load or a chronically overloaded drive keeps the coupling at high slip, all of that power becoming heat. The oil cooks, degrades, and the fusible plug blows — by design, to save the housing. Repeated near-stall starting without adequate cooling shortens oil life dramatically.
• Oil degradation and seal leakage. Heat oxidises the oil and hardens the shaft seals. A leak lowers the fill, which lowers torque capacity, which raises slip, which raises heat — a runaway. Routine oil analysis and seal inspection catch this early.
• Cavitation and vane erosion. At high slip and high circulation rate, local pressure in the vortex can drop below the oil's vapour pressure; vapour bubbles form and collapse against the vanes, pitting them over time. More common in aggressive variable-fill duty.
• Incorrect fill. Overfilling raises start torque and can defeat the soft-start benefit and overheat the motor; underfilling raises slip and runs the coupling hot. The fill quantity is a design parameter, not a top-up-to-the-line convenience.
• Fusible plug omission or wrong rating. Running a stall-prone drive with the wrong melt temperature, or a plug deliberately blanked off, removes the last line of defence and invites a burst housing or a fire.
• Bearing and shaft-seal wear. The only true contact surfaces. They set the maintenance interval; the blades themselves, bathed in oil, effectively never wear.

Common pitfalls when applying fluid couplings

• Expecting torque multiplication. A plain coupling is strictly 1:1. If you need launch torque boost, you need a torque converter with a stator, not a coupling.
• Sizing on torque alone. For stall-prone or frequent-start duty, heat capacity governs. Size for the thermal load of the worst-case start cycle, not just the running torque.
• Forgetting the permanent slip in speed calculations. Output speed is always a few percent below input. A pump or fan sized assuming 1:1 will run slightly slow; the 2–4 percent matters in precise flow applications.
• Ignoring cooling for high-cycle starts. A coupling that soft-starts perfectly once may overheat if started every few minutes. Add a delay chamber or external cooling for high duty cycles.
• Treating fill level as casual maintenance. Fill is a tuned design value. Top it up to the wrong level and you change the entire start-torque and slip behaviour.