Automobile Clutch Design: Torque Capacity, Types and Selection

The torque an automobile clutch can hold is set by one relationship, T = n x mu x W x Rm, where n is the number of friction surfaces, mu the coefficient of friction of the facing, W the axial spring load and Rm the mean friction radius. Everything else about clutch design follows from it: how big the plate has to be, how stiff the diaphragm spring is, how many plates a wet clutch needs, and how heavy the pedal ends up. A car clutch is normally sized to hold 1.4 to 1.8 times the peak torque of its engine.

This page is the design treatment: where the equation comes from, which pressure assumption designers use, a worked example with real numbers, and how the type and the actuation are chosen. If you want the general explainer instead, read what is a clutch, its types, applications and working.

The function of a clutch in an automobile

The clutch is the friction coupling between the engine flywheel and the gearbox input shaft, and it exists because an engine cannot be loaded from rest while the wheels start at zero. It does two jobs. It breaks the drive completely so a gear can be selected, since neither a sliding mesh gear box nor a modern synchromesh gear box will accept a gear while it is carrying torque. And it slips in a controlled way so the vehicle can move off while the engine keeps turning above its stall speed. Two secondary duties matter to the designer: the clutch acts as a torsional fuse that slips before something in the driveline breaks, and the damper springs in the plate hub absorb engine firing pulses so they never reach the gear teeth. The full walk-through of why a vehicle needs one is on the what is a clutch page.

Torque capacity of a clutch, derived

Consider one friction surface: an annulus of outer radius Ro and inner radius Ri, pressed by an axial load W, with coefficient of friction mu. Take a thin ring of radius r and width dr. Its area is 2 x pi x r x dr, the normal force on it is p x 2 x pi x r x dr, the friction force is mu times that, and the torque it contributes is that force multiplied by r:

dT = 2 x pi x mu x p x r² x dr, integrated from Ri to Ro, with W = 2 x pi x integral of p x r x dr over the same limits.

Both integrals need to know how the pressure p is distributed across the face, and that is where the two classical assumptions come in. Multiply the result by n, the number of rubbing faces, and you have the capacity of the whole assembly.

Uniform pressure

Assume p is the same everywhere on the facing. Then p = W / (pi x (Ro² – Ri²)), and carrying out the integration gives an effective mean radius of

Rm = (2/3) x (Ro³ – Ri³) / (Ro² – Ri²)

This describes a brand new clutch with perfectly flat, unbedded facings, and it is the condition a rigid, well-machined plate starts life in.

Uniform wear

Wear rate at any radius is proportional to pressure times rubbing velocity, and rubbing velocity is proportional to radius. For the facing to wear down evenly across its width, the product p x r must therefore be constant, which means pressure is highest at the inner radius and falls outwards. Integrating on that basis gives the much simpler

Rm = (Ro + Ri) / 2

and a maximum pressure at the inner edge of p_max = W / (2 x pi x Ri x (Ro – Ri)). That second expression is the one to use when checking the facing against its allowable pressure, because the inner edge is where the facing is worked hardest.

Which assumption designers use, and why

Uniform wear, almost always, for two reasons.

• It describes the real component. A clutch settles into the uniform wear condition within the first few hundred engagements and then stays there for the rest of its life. Uniform pressure describes it only while it is new.
• It is the conservative answer. Because pressure is redistributed towards the smaller radii, uniform wear always produces the lower mean radius and therefore the lower torque capacity. Designing on the higher number would mean a clutch that passes on paper and slips in service.

The uniform pressure result is kept as an upper bound, and for test-bench work on new plates. One useful consequence of the uniform wear analysis: for a given maximum allowable facing pressure and a given outer radius, torque capacity is greatest when Ri = Ro / square root of 3, that is about 0.577 Ro. Production clutches sit a little above that optimum, trading a small amount of torque for extra facing area and therefore extra heat capacity.

The number of friction surfaces, n

The term n is the number of pairs of rubbing faces, not the number of plates, and this is where most exam marks are lost.

• A single plate clutch has n = 2. The driven plate is faced on both sides, so it rubs on the flywheel on one side and on the pressure plate on the other.
• A twin plate dry clutch has n = 4.
• For a multi-plate stack with n1 driving discs and n2 driven discs, n = n1 + n2 – 1.

Because n multiplies directly, adding plates is the cheapest way to buy torque capacity without increasing diameter. That single fact is the whole design case for multi-plate clutches.

Worked example: sizing a single plate clutch

Take a clutch for a 1.5 litre petrol engine peaking at 145 Nm.

• Facing outside diameter 240 mm, so Ro = 0.120 m
• Facing inside diameter 160 mm, so Ri = 0.080 m
• Coefficient of friction of the organic facing, dry, mu = 0.30
• Total axial clamp load from the diaphragm spring, W = 4,000 N
• Single plate, faced both sides, so n = 2

Step 1, mean radius on the uniform wear basis.
Rm = (Ro + Ri) / 2 = (0.120 + 0.080) / 2 = 0.100 m

Step 2, torque capacity.
T = n x mu x W x Rm = 2 x 0.30 x 4,000 x 0.100 = 240 Nm

Step 3, check it against the engine.
240 / 145 = 1.655. That sits inside the usual 1.4 to 1.8 design band, so the clutch holds peak torque without slipping and still releases with a sensible pedal load. A factor much below 1.3 slips in service; much above 2.0 means an unnecessarily heavy pedal and a harsh take-up.

Step 4, check the facing pressure.
On the uniform wear basis the pressure peaks at the inner edge:
p_max = W / (2 x pi x Ri x (Ro – Ri)) = 4,000 / (2 x pi x 0.080 x 0.040) = 4,000 / 0.02011 = 0.199 MPa
That is inside the 0.1 to 0.3 MPa band for organic facings, so the design stands. Facing area per side is pi x (0.120² – 0.080²) = 0.02513 m², about 251 cm², which is what has to absorb the slipping heat.

Step 5, what uniform pressure would have given.
Rm = (2/3) x (0.120³ – 0.080³) / (0.120² – 0.080²) = (2/3) x (0.001728 – 0.000512) / (0.0144 – 0.0064) = (2/3) x 0.001216 / 0.008 = (2/3) x 0.152 = 0.101333 m
T = 2 x 0.30 x 4,000 x 0.101333 = 243.2 Nm, about 1.33 per cent higher than the uniform wear answer. The gap widens as Ri / Ro gets smaller, which is exactly when the conservative assumption matters most.

Step 6, the effect of a second plate.
Change nothing but the number of plates. The same facings, the same spring load and the same mu in a twin plate dry clutch give n = 4 and T = 4 x 0.30 x 4,000 x 0.100 = 480 Nm, in the same diameter.

Note what the equation does not contain: engine speed, vehicle weight and gradient are all absent. Capacity is a static friction question. Slipping time and heat are a separate calculation, and they are what actually kills clutches.

Requirements of a good clutch

• Torque capacity with a margin. Peak engine torque transmitted without slip, with a design factor of roughly 1.4 to 1.8.
• Gradual engagement. Take-up progressive over the pedal travel rather than on-off, which is the job of the waved cushion segments between the facings.
• Complete release. With the pedal down the driven plate must be genuinely free, or the gearbox drags and the gears crunch.
• Heat dissipation. Every joule lost while slipping becomes heat in the flywheel, plate and pressure plate. A clutch that cannot shed it glazes and fades.
• Low inertia on the driven side. A light driven plate stops spinning quickly when released, which is what lets the gearbox synchronisers work.
• Dynamic balance. It runs at engine speed, so imbalance reaches the whole car as vibration.
• Reasonable pedal effort. Around 100 to 150 N at the pedal for a passenger car, which is what governs the spring type and the linkage ratio.
• Even, slow wear and no sticking. Facings should wear evenly and the release mechanism should not need frequent adjustment.

The first three pull against each other, and most clutch design is the compromise between them. Raising W buys torque capacity and costs pedal effort; raising Rm buys torque capacity and costs inertia and space.

The parts that set the design numbers

Four of the parts appear directly in the torque equation. The flywheel and the pressure plate are the two driving faces that give a single plate clutch its n = 2, and both double as the thermal mass. The friction plate carries the facings that set mu and the radii that set Rm, and its splined hub is the only connection to the gearbox. The diaphragm spring, or the ring of coil springs in older designs, supplies W. The rest of the assembly exists to deliver or interrupt that load: the clutch cover carrying the spring, the release bearing that pushes the diaphragm fingers, the fork and linkage, and the spigot bearing in the crankshaft nose that keeps the plate concentric.

Facing material is the term most easily changed on a datasheet. Modern facings are woven or moulded composites of glass, aramid and brass fibres in a resin binder, giving a dry coefficient of friction of roughly 0.30 to 0.40 and an allowable contact pressure of about 0.1 to 0.3 MPa. Asbestos, once standard, is long gone. Sintered bronze or ceramic pucks lift mu to around 0.4 and tolerate far higher temperature, at the cost of harsh take-up, which is why they belong on competition and heavy-duty vehicles. For a description of each part and what it looks like, see the what is a clutch page.

How engagement happens, and where the heat goes

With the pedal up, the springs clamp the pressure plate against the driven plate and the driven plate against the flywheel, so the sandwich turns as one unit with no relative motion and no wear. With the pedal down, the release bearing pushes the diaphragm fingers, the diaphragm pivots on its fulcrum rings and its outer rim lifts, retracting straps pull the pressure plate back, and the clamp load falls to zero. Between those two states is the one that concerns the designer. The plate transmits partial torque while turning slower than the flywheel, and that speed difference multiplied by the transmitted torque is power, all of which becomes heat in roughly 251 cm² of facing. A hard start from rest can put tens of kilojoules in over a couple of seconds. This is why heat capacity, not torque capacity, usually decides the diameter on a heavy vehicle, and why riding the clutch destroys it. The step-by-step working of engagement is covered on the what is a clutch page.

Choosing a clutch type: duty, capacity and heat

Type selection is not a matter of taste. It follows from three questions: how much torque, how much slipping heat, and whether a driver is operating it at all.

Single plate: where the diameter limit bites

With n fixed at 2, the only ways to raise capacity are a larger Rm or a larger W, and both run into limits. Diameter is capped by the bellhousing and by the inertia penalty, since a bigger plate takes longer to stop spinning when released and makes the gearbox harder to shift. Clamp load is capped by pedal effort. That is why the single plate clutch runs out of capacity somewhere around 400 to 450 Nm in a car-sized package and why high-torque applications go to two plates rather than one larger one.

Its advantages are release quality and cooling. One plate between two faces separates cleanly, so drag on the gearbox input shaft with the pedal down is small, which is exactly what a synchromesh box needs. And the facing sits against two large iron masses that soak up slipping heat. Those two properties, not capacity, are why it remains the standard road car clutch.

Multi-plate, wet and dry: buying capacity with n

Stacking alternating driving and driven discs raises n directly, so torque scales with plate count at constant diameter. A racing 5.5 inch triple-plate unit holds more than a 9 inch single plate while carrying a fraction of the rotating inertia, which is worth more on a race car than the harsh take-up it costs.

The wet and dry choice is a trade between mu and heat. Running the stack in oil drops the coefficient of friction to roughly a third of the dry value, so a wet clutch needs several times as many surfaces for the same torque, and it drags slightly even when released. In exchange the oil removes heat continuously, which is why motorcycles, tractor PTO clutches and the clutch packs inside automatics and dual clutch units are nearly all wet: they slip repeatedly and a dry facing would not survive the duty. Dry multi-plate clutches, which Ducati used on many models, take the friction back and accept the heat and the rattle.

Centrifugal: when no driver is available

This type is selected when there is to be no clutch control at all. W is produced by engine speed rather than by a spring or a pedal: weighted shoes are held in by return springs at idle and are thrown outwards against a drum as speed rises, so engagement begins at a threshold set by the spring rate and the shoe mass.

That fixes the duty. The engagement speed cannot be varied by the operator, so the designer has to place it above idle but below the speed of peak torque, and everything else follows. The clutch cannot be held disengaged for a hill start or to coast, and it slips for longer on take-up than a foot-operated clutch does, which puts the heat limit rather than the torque limit in charge. Scooters, karts, chainsaws and small generator sets accept all of that in return for a machine that is driven with a throttle and a brake alone.

Semi-centrifugal: a light pedal on a heavy engine

The semi-centrifugal clutch exists to solve one specific design conflict, and it is worth understanding because the conflict itself never went away.

In a plain coil spring clutch, W has to be high enough to hold peak engine torque, and peak torque arrives at high engine speed. But the same W has to be pushed off the plate by the driver’s left foot at every junction and in every traffic queue. On a large, high-torque engine of the 1930s to 1960s, sizing the springs for peak torque produced a pedal effort well beyond what a driver would tolerate. Sizing them for a comfortable pedal produced a clutch that slipped.

The semi-centrifugal answer is to make part of the clamp load speed-dependent. The coil springs are deliberately made weaker, sized only for the torque the engine produces at low speed, where the driver actually operates the pedal. Bob weights are then mounted on the outer ends of the release levers. As engine speed rises, centrifugal force on those weights swings the levers so that they add clamping force to the pressure plate, on top of the spring load. Clamp load therefore rises with engine speed, which is roughly how engine torque behaves, so the clutch is strongest exactly when it needs to be.

The design costs are real: extra moving parts on a component turning at engine speed, more to wear at the lever pivots, more to go out of balance, and a clamp load that is now a function of two things rather than one, which makes the assembly harder to test and to set up. Wear on the lever pivots changes the weight geometry and therefore the added load, so the behaviour drifts over the clutch’s life.

What made it obsolete was not a better version of the same idea but a different spring. The diaphragm spring has a non-linear, near-flat load-deflection characteristic over its working range, which means it can supply a high clamp load while still requiring only a moderate release force, and it also holds clamp load nearly constant as the facing wears thin. That delivers what the semi-centrifugal design was reaching for, using one pressed component instead of springs plus weighted levers. Semi-centrifugal clutches survive today mainly on older vehicles and in some heavy-duty and agricultural applications where coil spring covers are still standard.

Clutch actuation: cable or hydraulic

Either system needs about 5 to 10 mm of free play at the pedal, so that the ball bearing in the release unit is not resting against the diaphragm fingers when the pedal is up. Lose that free play and two things follow at once: the bearing spins continuously and fails early, and a fraction of the clamp load is permanently held off the plate, so the clutch slips.

The automatic equivalents

Hydraulic torque converter. A conventional automatic replaces friction with fluid, and gets torque multiplication as a bonus. An impeller driven by the engine throws oil outwards into a turbine connected to the gearbox, and a stator between them redirects the returning oil so that it helps rather than opposes the impeller. That redirection is what multiplies torque, typically by around 1.8 to 2.5 to 1 at stall, falling to 1:1 as the turbine catches the impeller. There is no wear surface, so it can slip indefinitely, which removes the heat limit that governs friction clutch design. The cost is a standing efficiency loss, which is why modern converters carry a lock-up clutch that mechanically ties impeller to turbine once cruising.

Dual clutch transmission. A DCT is two ordinary friction clutches side by side, driving concentric input shafts, one carrying the odd gears and one the even. While the car runs in third, fourth is already engaged on the idle shaft, and the change is made by releasing one clutch as the other takes up, so torque is never fully interrupted. Sizing follows exactly the same T = n x mu x W x Rm rule as a manual clutch, applied twice. Medium and high torque units use wet multi-plate packs because the heat duty is high and repeated; dry dual clutches appear on lower-torque cars where the drag loss of running packs in oil costs more than it saves.

Diagnosing clutch problems

Work from the symptom to the cheapest check first. Most of these can be settled without removing the gearbox.

The engine revs climb but the car does not accelerate

That is slipping, and it shows up first in top gear on a gradient. Check free play at the pedal before anything else, because a lost free play adjustment holds part of the clamp load off the plate and mimics a worn clutch exactly. If free play is correct, look for oil: a leaking crankshaft rear seal or gearbox input shaft seal puts oil on the facing and collapses the effective mu, and a dry bellhousing drain hole is a useful clue. Only after those two does worn or glazed facing, or a weakened diaphragm spring, become the likely answer. Slipping generates heat quickly and the damage compounds, so it should not be driven on.

The car shakes fore and aft as the pedal comes up

Judder is almost never the pressure plate. Check the engine and gearbox mountings first, since worn mountings let the whole unit rock and turn a normal take-up into a shudder. Then look at the flywheel friction face for scoring, heat cracking or distortion, which is why a clutch change should include resurfacing or replacing the flywheel. Inside the plate itself, broken cushion segments or a distorted disc remove the progressive take-up, and oil contamination causes judder before it causes outright slip.

The gears crunch going into first or reverse

That is drag, meaning the plate is not fully released. Too much free play or air in the hydraulic line is the first thing to check, since both shorten the effective release travel. If the release travel is correct, the plate is being held: a seized spigot bearing in the crankshaft nose, a hub binding on rusted or worn input shaft splines, or a warped driven plate. On a wet multi-plate clutch, oil that is too thick or the wrong specification will also produce drag when cold.

A rumble or whine that appears when you press the pedal

That is the release bearing, loaded only while the pedal is down. Replace it with the clutch kit, since the gearbox has to come out either way. The mirror-image case matters more for diagnosis: a rattle present with the pedal up and the box in neutral, which disappears when you press the pedal, is the gearbox input shaft bearing, not the clutch at all. Pressing the pedal unloads that bearing and quietens it. Getting these two the wrong way round is the most common reason a clutch is replaced unnecessarily.

The pedal is heavy, or it stays on the floor

Heavy pedal with everything else normal usually means a corroded or badly routed cable, or a dry fork pivot, both of which are cheap to check before condemning the cover assembly. An uprated cover fitted to a standard engine will also give a heavy pedal, by design. A pedal that does not return at all is a snapped cable, a broken pedal return spring or a failed hydraulic seal, and the vehicle should not be driven.

References and further reading

• V. B. Bhandari, Design of Machine Elements, McGraw Hill, uniform pressure and uniform wear theories of friction clutches.
• Heinz Heisler, Advanced Vehicle Technology, Butterworth-Heinemann, friction clutch design, torque capacity and diaphragm spring characteristics.
• Kirpal Singh, Automobile Engineering, Volume 1, Standard Publishers Distributors, chapter on the clutch.
• AICTE Model Curriculum – Automobile Engineering.
• NPTEL – Automotive Systems lecture series, IIT.
• For the general explainer covering each part, the working cycle and the full list of clutch types, see what is a clutch. For the clutch in its place in the drivetrain, see parts of car transmission.

FAQs

How is the torque capacity of an automobile clutch calculated?

By T = n x mu x W x Rm, where n is the number of friction surfaces, mu the coefficient of friction, W the axial spring load and Rm the mean friction radius. Designers use the uniform wear assumption, which gives Rm = (Ro + Ri) / 2, because a clutch settles into that condition after bedding in and because it yields the lower, safer figure. For a 240 mm by 160 mm single plate clutch with mu = 0.30 and W = 4,000 N, Rm = 0.100 m and T = 2 x 0.30 x 4,000 x 0.100 = 240 Nm.

How many friction surfaces does a single plate clutch have?

Two. The driven plate carries friction facings on both of its faces, so one face rubs against the flywheel and the other against the pressure plate. That is why n = 2 in the torque equation for a single plate clutch, and using n = 1 is the most common mistake. A twin plate dry clutch has n = 4, and a multi-plate stack with n1 driving and n2 driven discs has n = n1 + n2 – 1.

What is the difference between uniform pressure and uniform wear in clutch design?

Uniform pressure assumes contact pressure is the same everywhere on the facing, which describes a new, unbedded clutch, and gives Rm = (2/3)(Ro cubed minus Ri cubed) / (Ro squared minus Ri squared). Uniform wear assumes the product of pressure and radius is constant, which is the condition a bedded-in facing reaches, and gives Rm = (Ro + Ri) / 2 with the highest pressure at the inner edge. Uniform wear always predicts the lower torque, so it is the design basis. For the worked example on this page the two differ by 1.33 per cent.

What is the function of a clutch in an automobile?

It connects and disconnects the engine from the gearbox. It breaks the drive so a gear can be selected, because a gear cannot be engaged while it is carrying torque, and it slips in a controlled way so the vehicle can start from rest while the engine keeps turning above its stall speed. It also works as a torsional fuse against driveline shock, and the damper springs in its hub keep engine firing pulses out of the gear teeth.

When is a multi-plate clutch used instead of a single plate clutch?

When the torque required exceeds what the available diameter and a tolerable pedal effort can deliver with n = 2. Because n multiplies torque capacity directly, adding discs raises capacity without increasing diameter or clamp load, so multi-plate units suit motorcycles, racing cars and high-torque drives where space and rotating inertia are limited. Running the stack in oil cuts mu to roughly a third of the dry value but removes heat continuously, which is why clutches that slip repeatedly, such as motorcycle and automatic transmission packs, are wet.

Related Topics on EngineeringHulk

• 👉 Clutch Friction Materials
• 👉 Synchromesh Gear Box
• 👉 Theory of Machines – Study Guide