Coupling Torque and Selection Calculator
Coupling Torque and Selection Calculator
Selection torque for a flexible coupling from power, speed and a published service factor — then checked against what the prime mover can actually deliver, which for an induction motor is usually the larger number. With misalignment allowances by coupling type, the rule that they cannot all be used at once, and the bearing reaction the coupling puts back into the shaft.
Coupling torque and selection
22 kW at 1,470 rev/min into an uneven-load machine running 8 hours a day, driven by a NEMA design B motor, through a straight-sided jaw coupling
Two criteria, and the larger governs
- K_s
- coupling service factor. A range, by load class, with an additive running-time adjustment. Different table from the chain and belt one and it goes nearly twice as high
- breakdown torque
- the most an induction motor can produce before it stalls: 175 to 300 per cent of rated for a NEMA design A or B. Not a fault condition — it happens on every hard start
- k_lateral
- the coupling’s lateral spring rate, which is what turns misalignment into a bearing load. From the maker’s data sheet; this page will not guess it
Worked example
22 kW at 1,470 rev/min into an uneven-load machine running 8 hours a day, driven by a NEMA design B motor, through a straight-sided jaw coupling
Nominal torque: 9,549 × 22 / 1,470 = 142.9 N·m. That is the average torque the coupling passes and it is the last time the average matters
Service factor: an uneven load at 8 hours a day is 1.5 to 2.5 on Cross+Morse's scheme, so the selection torque is 214 to 357 N·m. Note that the band is 1.67 times wide — the catalogue gives a range and expects you to place yourself in it
Now the peak. A NEMA design A or B motor's breakdown torque is 175 to 300 per cent of rated, so this motor can deliver 250 to 429 N·m — and it will, on a hard start or if the driven machine jams
The high end of the motor's own capability, 429 N·m, is 1.20 times the TOP of the service-factor band. So the peak governs and the coupling has to be rated for 429 N·m, not for the service factor's answer. That is the single most useful thing on this page and it is why a coupling chosen on power and speed alone fails on the third hard start rather than after a year
Misalignment: 0.3° of angle against the jaw coupling's 1° allowance is 0.300 of it; 0.15 mm of offset against 0.4 mm is 0.375; 0.5 mm of axial against 1.0 mm is 0.50. Individually all fine
Added together they come to 1.175, which is over one — and Lovejoy's instruction is explicit: “do not operate with both angular and parallel misalignment at their maximum values”. The three consume the same strain in the same rubber element. This alignment needs improving even though no single figure is over its allowance
The bearing reaction is left blank, deliberately. The mechanism is a lateral spring rate times the offset, and Machine Design say that vendors' catalogues rarely publish the rate. Enter it if your data sheet has it; if it does not, ask, because the load goes into your bearings whether or not anybody calculated it
The coupling service factor, and how it differs from the chain and belt one
| Load class | Running time | Factor, low | Factor, high | Selection torque at 22 kW and 1,470 rev/min (N·m) |
|---|---|---|---|---|
| Even load | Under 4 hours a day | 0.90 | 1.90 | 272 |
| Even load | About 8 hours a day | 1.00 | 2.00 | 286 |
| Even load | About 16 hours a day | 1.20 | 2.20 | 314 |
| Even load | 24 hours a day | 1.30 | 2.30 | 329 |
| Uneven load | Under 4 hours a day | 1.40 | 2.40 | 343 |
| Uneven load | About 8 hours a day | 1.50 | 2.50 | 357 |
| Uneven load | About 16 hours a day | 1.70 | 2.70 | 386 |
| Uneven load | 24 hours a day | 1.80 | 2.80 | 400 |
| Heavy shock | Under 4 hours a day | 2.40 | 3.10 | 443 |
| Heavy shock | About 8 hours a day | 2.50 | 3.20 | 457 |
| Heavy shock | About 16 hours a day | 2.70 | 3.40 | 486 |
| Heavy shock | 24 hours a day | 2.80 | 3.50 | 500 |
Peak torque: what the prime mover can actually deliver
| Prime mover | Breakdown torque, low | Breakdown torque, high | Peak at 22 kW and 1,470 rev/min (N·m) | Source |
|---|---|---|---|---|
| Electric motor, NEMA design A or B | 175% | 300% | 429 | NEMA MG 1 design letters, as L&B Electric summarise them |
| Electric motor, NEMA design C | 190% | 225% | 322 | NEMA MG 1 design letters, as L&B Electric summarise them |
| Electric motor, NEMA design D | 275% | 275% | 393 | NEMA MG 1 design letters, as L&B Electric summarise them |
| Diesel or petrol engine — see the note | — | — | — | Not quantified from a standard here — see the note below the calculator |
| Turbine or hydraulic motor, smooth | 110% | 150% | 214 | NEMA MG 1 design letters, as L&B Electric summarise them |
Published misalignment allowances by coupling type
| Coupling type | Angular (°) | Parallel (mm) | Axial (mm) | Angular in mrad | Where the figure comes from |
|---|---|---|---|---|---|
| Jaw / spider, straight-sided (Cross+Morse L, Lovejoy L) | 1.00 | 0.40 | 1.00 | 17.45 | Cross+Morse L series; Lovejoy give ½° to 1° and 0.010–0.015 in |
| Jaw / spider, curved-jaw | 1.30 | 0.50 | 1.70 | 22.69 | Lovejoy curved jaw 1.3° angular; Cross+Morse KE 1° / 0.5 mm / 1.7 mm |
| Shear-type rubber sleeve (“donut”) | 1.00 | 0.25 | 1.00 | 17.45 | Lovejoy: 1° angular, 0.010 in parallel at size 3 rising to 0.062 in at size 16 |
| Tyre / corded flexible | 4.00 | 3.20 | 1.00 | 69.81 | Lovejoy: up to 4° angular and up to 1/8 in parallel |
| Pin-and-bush (Cross+Morse GE) | 1.50 | 1.80 | 2.40 | 26.18 | Cross+Morse GE series |
| Rubber-in-compression, high flexibility (Morflex) | 5.00 | 1.25 | 1.00 | 87.27 | Cross+Morse Morflex |
| Gear coupling (Cross+Morse GFA/GFAS) | 1.00 | 0.80 | 2.00 | 17.45 | Cross+Morse GFA/GFAS series |
What a flexible coupling actually does about misalignment
| The belief | What happens | What to do |
|---|---|---|
| “A flexible coupling corrects misalignment” | It does not. It SURVIVES misalignment, by deforming once per revolution. The shafts stay exactly as misaligned as you left them, and the coupling converts that geometry into a cyclic strain in itself and a cyclic load in the two sets of bearings | Align the shafts. The coupling’s allowance is a tolerance on how badly you managed, not a substitute for doing it |
| “Within the catalogue allowance, misalignment is free” | It is not free, it is survivable. The element’s fatigue life falls with misalignment, and the reaction it pushes into the bearings rises in proportion to the offset. Machine Design put it bluntly: the reaction forces “may seem surprisingly high to engineers who never considered them before” | Treat the allowance as an endurance limit rather than a target. Half the allowance costs very little and buys a lot |
| “The allowance is a fixed property of the coupling” | It falls as the torque rises. An elastomeric element carrying rated torque is already strained and has less strain left to absorb misalignment, which is why catalogues quote the allowance at rated torque | If you need the full misalignment allowance, size the coupling above the torque so it is not also fully strained |
| “Angular and parallel allowances can both be used fully” | They cannot. Lovejoy state it directly: “do not operate with both angular and parallel misalignment at their maximum values”. The two consume the same strain in the same element | Add the utilisations, as this page does, and keep the total below one |
| “A single-element coupling handles offset” | It handles ANGLE. Pure parallel offset with no angle requires two flexing planes — which is why a jaw coupling, a beam coupling and a disc coupling with one element are all poor at offset and a double-disc or an Oldham is good at it | Count the flexing planes. One plane, one degree of angular freedom; two planes, offset as well |
AGMA coupling balance classes — a ladder, and no rpm threshold
| AGMA class | Maximum displacement of the principal inertia axis (rms micro-inches) |
|---|---|
| 4 | over 32,000 |
| 5 | 32,000 |
| 6 | 16,000 |
| 7 | 8,000 |
| 8 | 4,000 |
| 9 | 2,000 |
| 10 | 1,000 |
| 11 | 500 |
| 12 | 250 |
Why the service factor is usually the smaller number
The service factor is not the criterion. The prime mover usually is. Selection torque is nominal torque times a service factor, and for an uneven load running eight hours a day the published factor is 1.5 to 2.5. But the motor driving it can deliver 175 to 300 per cent of its rated torque at breakdown, and it will — on every hard start, and every time the driven machine jams. At the defaults on this page the motor’s own peak is above the top of the service-factor band, so the peak governs and the service factor is the smaller number. This page computes both and takes the larger, because a coupling chosen on the service factor alone fails on the third hard start rather than gradually.
And it is a different table from the chain and belt one. That matters because the two get confused. A coupling service factor runs to 3.2, and to 3.5 round the clock; the tabulated chain and V-belt families top out at 1.8. The reason is structural: a chain or belt drive has slip, catenary and elasticity between the shock and the component, while a coupling is a rigid torque path and takes the shock undamped. The duty handling differs too — the coupling table ADDS a running-time adjustment where the drive tables use separate columns. The drive service factor calculator on this site does the chain and belt case, with three standards families printed side by side; do not carry a number across.
A flexible coupling does not fix misalignment. It survives it, and it charges the bearings for the privilege. The shafts stay exactly as misaligned as you left them; the coupling converts that geometry into a cyclic strain in itself and a cyclic load in two sets of bearings, once per revolution, for the life of the machine. Machine Design observe that the reaction forces “may seem surprisingly high to engineers who never considered them before” and that the reaction “increases proportionally with shaft deflection”, so it can be “defined as a spring rate and expressed as force per unit deflection”. They also observe why nobody calculates it: “vendor’s catalogs rarely give coupling-induced reaction forces, consequently designers overlook them as a selection factor.” So this page computes F = k × offset and asks you for k. It will not invent one, because a plausible invented number gets used.
Three misalignments, one element, and they add. Angular, parallel and axial displacement all consume strain in the same flexible member, and each catalogue allowance is quoted as though the other two were zero. Lovejoy say it outright: “do not operate with both angular and parallel misalignment at their maximum values”. This page therefore adds the three utilisations and warns when the total passes one — which at the defaults here it does, even though no single figure is over its own limit. Two further points that catch people. The allowance FALLS as the torque rises, because an element carrying rated torque is already strained; catalogues quote the allowance at rated torque. And a single-element coupling accommodates ANGLE, not offset: pure parallel misalignment needs two flexing planes, which is why a jaw or a single-disc coupling is poor at offset and a double-disc or an Oldham is good at it.
What is refused. No balance-speed threshold, because ANSI/AGMA 9000-D11 does not give one and Ameridrives’ explanation of it says that “no generally applicable specific values can be given” — the class ladder is printed and the choice is left where the standard leaves it. No peak-torque multiplier for a reciprocating engine, because the relevant quantity is the engine’s cyclic irregularity and no general figure was sourced; the page names the problem instead, and points out that an engine drive’s real risk is torsional resonance rather than peak torque. No torsional stiffness selection and no torsional vibration analysis. And no bore or hub dimensions, which are the maker’s. For what the coupling’s own reaction does to the shaft, the shaft deflection and slope calculator takes a transverse load and gives the slope at the bearing; for whether the assembly is anywhere near a lateral resonance, the shaft critical speed calculator.
Frequently asked questions
How do I size a flexible coupling?
Take the nominal torque from the power and speed, multiply by the service factor for the driven machine’s load class and running hours, and then — the step that gets missed — check that result against what the prime mover can actually deliver at breakdown. For an ordinary induction motor that is 175 to 300 per cent of rated torque, which is usually the larger of the two numbers. Size for the larger. Then check the misalignment allowances, adding the three utilisations rather than checking them separately.
Is the coupling service factor the same as the chain or belt one?
No, and they should not be interchanged. Coupling factors run from 1.0 to about 3.2, and to 3.5 round the clock; the tabulated chain and V-belt families top out at 1.8. A coupling is a rigid torque path with nothing between the shock and the component, while a belt slips and a chain has catenary and both have elasticity, so the same machine class earns a bigger factor on a coupling. The coupling table also handles running hours by adding an increment rather than by separate columns.
Does a flexible coupling correct misalignment?
No. It survives misalignment. The shafts remain exactly as misaligned as they were, the coupling deforms once per revolution to accommodate the geometry, and the reaction goes into the bearings at each end. That reaction is proportional to the offset and behaves as a lateral spring rate, and it is the load most often left out of a bearing calculation — largely because makers rarely publish the rate. Align the shafts; the coupling’s allowance is a tolerance on how well you managed.
Can I use the full angular and the full parallel allowance at once?
No. Lovejoy’s handbook says explicitly not to operate with both at their maximum values, and the reason is that both consume strain in the same element while each published allowance assumes the other is zero. Adding the fractions and keeping the total below one is the simple defensible rule, and it is what this page does. At the defaults here the three utilisations total more than one even though every individual figure is comfortably inside its own allowance.
Which coupling type tolerates the most misalignment?
Of the types on this page, a rubber-in-compression or tyre coupling — 4 to 5 degrees of angular against 1 degree for a straight jaw. You pay for it in torsional softness, in element life, and in the fact that a very flexible coupling can move the drive’s torsional natural frequency into the excitation range. An Oldham or double-disc design is the answer for large PARALLEL offset specifically, because it has two flexing planes; a single-plane coupling of any type handles angle rather than offset.
At what speed does a coupling need balancing?
There is no such speed in the standard and this page will not invent one. ANSI/AGMA 9000-D11 classifies potential unbalance in classes 4 to 12, each halving the permitted displacement of the principal inertia axis, and the published explanation of it states that no generally applicable specific values can be given — the class comes from the machine’s sensitivity, its bearing loads, and how close it runs to a critical speed. The practical questions are what class the coupling comes in as standard, what the machine’s vibration specification allows, and whether there is a lateral resonance nearby.
What about a diesel engine drive?
Different problem. A reciprocating engine’s torque has strong harmonics at multiples of the firing frequency, so the peak instantaneous torque is not a fixed multiple of the mean and depends on the cylinder count and the engine’s own cyclic irregularity — the engine maker’s data, not a general table. More importantly, the real failure mode on an engine drive is torsional resonance rather than peak torque: the coupling’s torsional stiffness has to be chosen so the drive train’s torsional natural frequency avoids the engine’s excitation orders. That is a torsional vibration analysis and it is outside this page, which is why the peak figure is left blank for this prime mover.
Related calculators
References
- Cross+Morse. Shaft Couplings catalogue. The source for the coupling service-factor scheme printed here: even load 1.0 to 2.0, uneven load 1.5 to 2.5, heavy shock 2.5 to 3.2, all at 8 hours a day, with running-time adjustments of −0.1 below 4 hours, +0.2 at 16 hours and +0.3 at 24 hours; and for the per-series misalignment allowances (jaw 1° / 0.4 mm / 1.0 mm, GE 1.5° / 1.8 mm / 2.4 mm, Morflex 5° / 1.25 mm / 1.0 mm, gear 1° / 0.8 mm / 2.0 mm) and the maximum bores.
- Lovejoy (a Timken company). The Lovejoy Coupling Handbook. The source for the elastomeric misalignment allowances quoted here — straight-side jaw ½° to 1° angular and 0.010 to 0.015 in parallel, curved jaw 1.3° angular, shear-type sleeve 1° angular, corded tyre up to 4° angular and up to 1/8 in parallel — for the peak-torque capability of a jaw coupling in compression (“6 or 7 times the nominal rating”, against “only 3 or 4 times” for a shear-type), and for the instruction that matters most: “do not operate with both angular and parallel misalignment at their maximum values”.
- NEMA MG 1, Motors and Generators, design letters A, B, C and D. Cited by designation; not fetched. The torque characteristics quoted are from L&B Electric’s published summary: design A and B locked-rotor 70–275% and breakdown 175–300% of rated torque; design C locked-rotor 200–285% and breakdown 190–225%; design D 275% both, with 5–8% slip. The brief’s “2 to 3×” is right for the commonest motor there is and the range is wider than that at both ends.
- ANSI/AGMA 9000-D11, Flexible Couplings — Potential Unbalance Classification (the revision of AGMA 515.02). Cited by number; not reproduced. Ameridrives’ published explanation carries the class ladder — classes 4 to 12, each halving the permitted displacement of the principal inertia axis from over 32,000 down to 250 rms micro-inches — and, importantly, no rpm threshold: the class is chosen from the sensitivity of the machine, and “no generally applicable specific values can be given”. This page therefore prints the ladder and refuses to name a speed above which a coupling must be balanced.
- Machine Design. “How coupling types affect bearing forces.” The source for the mechanism this page computes and for the admission that nobody publishes the constant: the reaction “increases proportionally with shaft deflection” and can be “defined as a spring rate and expressed as force per unit deflection”, but “vendor’s catalogs rarely give coupling-induced reaction forces, consequently designers overlook them as a selection factor”. So the page takes the spring rate as an input and says where to get it.
- Ruland Manufacturing. Avoiding Coupling Failure — Considerations in Design Selection. Read for the type-by-type qualitative picture: a beam coupling “easily accommodates angular misalignment and axial motion with a lower capability to compensate for parallel misalignment”, an Oldham coupling is “well suited for handling relatively large amounts of parallel misalignment” with “low capability to compensate for angular”. Its own comparison chart rates every property as Low / Moderate / High rather than in degrees, so no number on this page comes from it.