# Keyway transfer forces

33 replies to this topic

### #1 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 27 February 2012 - 07:57

How does one find out what forces a keyway can transfer?

I found this calculator http://www.tribology...ulators/key.htm But it does not make much sense.

I have the tranfer torque.

Shaft dia = 75mm

arm length = 300mm

Mv = Force * arm length to the outer dim of the shaft.

Mv = 15000N * (300 -(75\2) = 3937500Nmm

Wp = (pi * 75^3)/16 = 82835mm^3

Tau twist = Tt = Nmm/mm^3 = 3937500/82835 = 47.5N/mm^2

So the shaft is good to go...

And why the hell did i never hear anything about this at school? it should be mandatory.. such a basic thing.

Keyway dim: 22 wide 9 deep on shaft.

Edited by MatsNorway, 27 February 2012 - 08:00.

### #2 Wolf

Wolf
• Member

• 7,881 posts
• Joined: June 00

Posted 27 February 2012 - 11:25

How does one find out what forces a keyway can transfer?

I found this calculator http://www.tribology...ulators/key.htm But it does not make much sense.

I have the tranfer torque.

Shaft dia = 75mm

arm length = 300mm

Mv = Force * arm length to the outer dim of the shaft.

Mv = 15000N * (300 -(75\2) = 3937500Nmm

Wp = (pi * 75^3)/16 = 82835mm^3

Tau twist = Tt = Nmm/mm^3 = 3937500/82835 = 47.5N/mm^2

So the shaft is good to go...

And why the hell did i never hear anything about this at school? it should be mandatory.. such a basic thing.

Keyway dim: 22 wide 9 deep on shaft.

IIRC, the key is in a simplified consideration loaded in shear (force over key cross section surface width*length), but primary concern for dimensioning is contact pressure... It transmits force by half of it's total height (contact stress is force over half of the area height*length), but actually the keyway is almost always set deeper in shaft that in outer element. If you read german, I could post a details of the calculation... HTH

I think you can download MIT Calculator for such things- ISTR that it has a 30 day free trial, it binds into Excel, and does amazing number of tasks such as this, spline calculations, gears, belts and chains, &c...

### #3 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 27 February 2012 - 14:04

So this is good to go then as long as i have better than S355 steel.

http://i1188.photobu...pg?t=1330351314

Im actually considering stainless because it is going below ground.

### #4 Wolf

Wolf
• Member

• 7,881 posts
• Joined: June 00

Posted 27 February 2012 - 14:22

So this is good to go then as long as i have better than S355 steel.

http://i1188.photobu...pg?t=1330351314

Im actually considering stainless because it is going below ground.

Mats, I think the contact pressure will have little to do with type of steel used... The book I'm quoting here says it's 120-180MPa for steels (oscillating force is supposed to reduce those numbers to 70%, and if load changes direction to 50%). Here are enclosed standard key and keyway dims, tolerances as well as preferred lengths and calculations. HTH

### #5 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 27 February 2012 - 18:16

Mats, I think the contact pressure will have little to do with type of steel used...

I know that.... its relevant for the safety margin...

And this part is not rotating. its only moving about 30-45 degrees.

The Key dim is 22 x 14mm about 150mm long.

### #6 Greg Locock

Greg Locock
• Member

• 4,854 posts
• Joined: March 03

Posted 03 March 2012 - 10:50

Incidentally, I know people do it, but a keyway should be used to provide an angular constraint for assembly, not react torques. The reliable way to transfer torques is tapers or interference fits or loctite, followed by splines, followed by any number of horrible devices, including keyways.

### #7 malbear

malbear
• Member

• 246 posts
• Joined: September 02

Posted 18 March 2012 - 10:40

Incidentally, I know people do it, but a keyway should be used to provide an angular constraint for assembly, not react torques. The reliable way to transfer torques is tapers or interference fits or loctite, followed by splines, followed by any number of horrible devices, including keyways.

so a taper with a keyway and a thread with castle nut with split pin would be the best to hold an arm position onto a shaft. for example a steering arm out of a box.

### #8 Canuck

Canuck
• Member

• 1,818 posts
• Joined: March 05

Posted 18 March 2012 - 15:57

Incidentally, I know people do it, but a keyway should be used to provide an angular constraint for assembly, not react torques. The reliable way to transfer torques is tapers or interference fits or loctite, followed by splines, followed by any number of horrible devices, including keyways.

Interesting. I would have thought the greater surface area of a spline would be preferable to a taper. Having said that Harley-Davidson crankshafts used to be 5-piece affairs: sprocket shaft, flywheel, crankpin, flywheel, pinion shaft. The sprocket and pinion shafts as well as one end of the crank pin were all taper/key assemblies while the remaining end of the crankpin was simply taper and all were backup up with a nut. It's been a while since I've been into one, but it seems to me the current iteration is three-pieces (flywheel/pinion shaft and flywheel/sprocket shaft as single machined castings and crank pin) that is pressed together in a 50-ton press. No tapers, no nuts (there's a Harley rider joke sitting there...).

### #9 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 18 March 2012 - 20:55

I have simplified the design so that a light shrink fit/pressfit will be possible. I posted the old design in the rolling friction thread.
http://forums.autosp...howtopic=161861

Now that i have the center piece in the picture above integrated with one of the stoppers i no longer have anything beond(inside of) the bearing houses. Now i dare to use a shrink fit.

Main issue in all of this now is to create the speed dampers.. i don`t want to reduce the slope as that would make it easier for the wheels to become stuck on small stones/ice or just rubbing against the inside rail flange.

Splines is the prefered method for train stabilisers. apart from that odd design with shrink fit that needs to be replaced in one piece(pulling of a axle thread)
http://forums.autosp...howtopic=156684

Picture later if of interest.

Edited by MatsNorway, 18 March 2012 - 21:04.

### #10 johnny yuma

johnny yuma
• Member

• 928 posts
• Joined: May 07

Posted 20 March 2012 - 05:17

so a taper with a keyway and a thread with castle nut with split pin would be the best to hold an arm position onto a shaft. for example a steering arm out of a box.

My first car a 1953 Holden had the hub,taper and keyway arrangement for the rear axle.It began to creak,so I took the car to Self Help Engineering in Waitara,Sydney where a gruff German named Deiter had a workshop where you used his facilities and he offered help. I told him no matter how much I tightened the centre nut the hub creaking remained.He replied you must resurface the taper and hub... "the taper is not there for a joke !"(quote) Do you really think the keyway drives the wheel ! (you stupid boy) .Here is the grinding paste do it until you are too tired to continue...Hey it worked of course !

### #11 fredeuce

fredeuce
• Member

• 368 posts
• Joined: May 07

Posted 20 March 2012 - 21:55

When I went to Trade School the way they explained the working of the tapered joint on a axle was that if you assembled it properly you could drive it without the key being present.As stated by Greg it is only there to locate everyting . The taper does the work.

Why can't you weld those arms on the end of the shaft ? Or the one in the middle for that matter? You can still remove the bearing caps to service if need be.

### #12 Greg Locock

Greg Locock
• Member

• 4,854 posts
• Joined: March 03

Posted 20 March 2012 - 22:06

If you draw a cross section through a keyed joint you'll see that if the taper is loose, and the drive is via the keyway, the brunt of the force is taken by the edges of the cutouts of the keyway, and a rather nasty shearing/bending action on the key itself. This is actually unstable, the forces are working to drive the joint apart.

The saving grace is that as the key falls over and wedges the joint apart, the shaft and wheel on the opposite side are forced together, allowing some of the torque to be transferred by friction.

A taper is better than a spline because there are no stress raisers at all, so the fatigue life is excellent. The system is also self centreing. A spline is better because it can slide axially, when unloaded, but the bending of the splines reduces the fatigue life. OTOH a rolled spline has excellent mechanical properties in its own right. Centreing is also a bit more haphazard with a spline, unless you use end contact on the teeth, which various people have tried as a solution. Gear teeth are designed to contact on the flanks, not the tips and roots.

### #13 bigleagueslider

bigleagueslider
• Member

• 1,106 posts
• Joined: March 11

Posted 21 March 2012 - 06:22

If you draw a cross section through a keyed joint you'll see that if the taper is loose, and the drive is via the keyway, the brunt of the force is taken by the edges of the cutouts of the keyway, and a rather nasty shearing/bending action on the key itself. This is actually unstable, the forces are working to drive the joint apart.

The saving grace is that as the key falls over and wedges the joint apart, the shaft and wheel on the opposite side are forced together, allowing some of the torque to be transferred by friction.

A taper is better than a spline because there are no stress raisers at all, so the fatigue life is excellent. The system is also self centreing. A spline is better because it can slide axially, when unloaded, but the bending of the splines reduces the fatigue life. OTOH a rolled spline has excellent mechanical properties in its own right. Centreing is also a bit more haphazard with a spline, unless you use end contact on the teeth, which various people have tried as a solution. Gear teeth are designed to contact on the flanks, not the tips and roots.

Greg Locock,

You make some excellent points regarding the merits of transferring torque across a shaft joint using friction instead of shear in a key. The OP asked about the forces a keyway can transfer. As you noted, the problem is much more complex than simple shear thru the key body. You must consider relative structural stiffnesses, stress concentrations, fits and clearances, load reversals, etc. With shaft joints, if there is even the slightest amount of relative movement at the joint interface, fretting will likely occur. It is virtually impossible to create shaft joints that have no relative motion relying on keyways alone. Keyways are suitable as indexing devices, but are not a good choice for transferring torque.

Friction joints (tapers, shrink fits, hydraulic fits, clamped fits, etc.) are the only sure way to prevent fretting. Each approach has its own benefits and drawbacks in terms of cost or ease of assembly.

slider

### #14 Magoo

Magoo
• Member

• 2,683 posts
• Joined: October 10

Posted 21 March 2012 - 10:28

All those right angles and sharp edges in a key/keyway scare me to death.

Crank-driven supercharger setups are rather prone, naturally enough, to breaking crankshafts in the region of the drive pulley. Of course, the pulley is generally secured to the crank with a key. When you take the broken pieces apart, guess where the fracture originates. Every single time.

### #15 cheapracer

cheapracer
• Member

• 10,388 posts
• Joined: May 07

Posted 21 March 2012 - 13:21

All those right angles and sharp edges in a key/keyway scare me to death.

Alas, there is plenty of straight shaft pulleys in industry that the key takes the whole brunt in shear.

My own 3 die tube roller is an example, generally where pulleys or dies are changed often of course.

### #16 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 21 March 2012 - 20:36

Its not a race car!

It will have a slight shrinkfit. Probably

and im working on speed dampers to reduce the impact from wheels going all the way from the top.

Those are the main issue now i feel.
I have not idea about how to calculate the impact forces on that one.

And odly they spesify the max Newtons on the dampers while impact forces is all about joules supposedly.

Most likely im just going to oversize everything get it mounted.

### #17 hogits2

hogits2
• Member

• 68 posts
• Joined: March 10

Posted 21 March 2012 - 21:44

Not trying to be awkward but how about extending both the arms down and putting a cylinder on each one. That would reduce the load on keys etc.

Not sure if I understand this - is it just raising a bar to stop the loose axles rolling past? How will you unload it to release it? If you are just raising a bar ( or 2 uprights) why not use a fabricated piece with a simple pivot at each bottom corner. H frame or similar. No keyways needed.

Or a rotating X frame to allow axles past one at a time - like a turnstile on it's side.

Or....

### #18 Greg Locock

Greg Locock
• Member

• 4,854 posts
• Joined: March 03

Posted 21 March 2012 - 22:26

I finally got back to my desk

SAE formula for torque capaicty of a straight (not involute) spline =1000*number of splines*mean radius*spline depth*Length all in lb and in

torque capacity of a properly fitted soft key is the shear strength of the key, divide by FoS, which varies from 1.5 for smooth running to 4.5 for reversing impact torques

A less lazy person than me could look at those two equations, and compare it with the torque capacity of a taper or interference fit, which is .187*mean radius*area of taper*contact pressure, divided by FoS which are going to be similar to those for a key.

Hard and tapered keys will produce results somewhere between a taper and a soft key.

Best books for learning mechanical design are usually by Shigley, but this is from deutschman. michels and wilson, which is bit less of a textbook and a bit more of a design manual.

### #19 cheapracer

cheapracer
• Member

• 10,388 posts
• Joined: May 07

Posted 22 March 2012 - 01:28

Small world Mats, I was playing with trains on the weekend just gone ...

... had a meeting with a man who was loading 85 of those yellow iron ore wagons bound for Oz. They use an air ram as well to open the lower chutes. I went to the factory where the wagons were made but didn't have time to get to the locomotive factory sad to say.

The ships side mounted cranes put them into the holds, I was amazed at how much 20 tons rocked a 13,000 tonne ship as they were being picked up from the wharf, this pic from the ship's bridge, the tarped wagons were going on deck and those red chassis were off to Angola (very narrow gage too), look carefully and you can see iron ore wagons in the hold..

One more for fun, this crane was putting the wagons on the truck you see above and coming back around to the side of the ship, they were loading a wagon every 4 minutes.

Edited by cheapracer, 22 March 2012 - 01:46.

### #20 Catalina Park

Catalina Park
• Member

• 6,189 posts
• Joined: July 01

Posted 22 March 2012 - 06:45

My brother used to play with train axles a few years back. He explained to me how they would press the old wheels off the axle and then crack test the axles and then fit new wheels. He made it all sound easy.

### #21 Lee Nicolle

Lee Nicolle
• Member

• 7,272 posts
• Joined: July 08

Posted 22 March 2012 - 09:17

All those right angles and sharp edges in a key/keyway scare me to death.

Crank-driven supercharger setups are rather prone, naturally enough, to breaking crankshafts in the region of the drive pulley. Of course, the pulley is generally secured to the crank with a key. When you take the broken pieces apart, guess where the fracture originates. Every single time.

I agree with that 100%. I have seen them severely cracked just driving a dry sump pump [and watewrpump] in that area.
Though Mats is playing with tapers with keys which does seem to be stronger.And still not my favorite piece of engineering though it has been in use forever.
Johnnys early Holdens with that set up is a example, similar on Studebakers did break the ends off the axles though that seems an anomoly. but coming loose and stripping the key is so common. Ford V8s do that too.

### #22 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 22 March 2012 - 16:00

Not trying to be awkward but how about extending both the arms down and putting a cylinder on each one. That would reduce the load on keys etc.

Not sure if I understand this - is it just raising a bar to stop the loose axles rolling past? How will you unload it to release it? If you are just raising a bar ( or 2 uprights) why not use a fabricated piece with a simple pivot at each bottom corner. H frame or similar. No keyways needed.

Or a rotating X frame to allow axles past one at a time - like a turnstile on it's side.

Or....

Having a sylinder on each end is absolutely a option, Main issue is clearance to the wall. and double everything -> price.

Unloading happens with the sylinder, stopping gets done with a spring. its a beefy versjon of a single acting sylinder. i want this as a safety to stop wheels from rolling about when something pnumatic/electric fails.

What you mean pivot? ball joint? does SKF for instance have such units? i dunno where to get such btw.

I was originally thinking brass bearings and lubrication nipples but SKF had the ball bearing stuff ready so faster and easier to take those.. and safer.

Right now this is on "hold" as i have desided to focus on a new customer that i hope to see more of. And that tool/unit is a fairly demanding design. Everything must be clamped with quick releases and must be fairly light. All aluminium frame.

### #23 malbear

malbear
• Member

• 246 posts
• Joined: September 02

Posted 05 April 2012 - 22:22

This is how we machine a keyway in a 250kg shaft in the Ausy bush.
mal

Edited by malbear, 05 April 2012 - 22:32.

### #24 bigleagueslider

bigleagueslider
• Member

• 1,106 posts
• Joined: March 11

Posted 07 April 2012 - 02:24

I design aircraft turbine engine drivetrains. We never use tapers or keys for shaft connections. We do use curvics, involute splines, and bolted flanges. The bolted flanges are designed to transmit the full torque load (plus a 1.5 FoS) using friction alone. The curvics are designed to transmit the same torque load with sufficient preload to prevent separation. The splines are sized to transmit the full torque load based on a flank contact stress (P/A) limit of <5 ksi.

Take a look at a typical turbine engine bolted flange joint. They are designed to transmit torques using friction alone, and that's why they use large numbers of bolts.

slider

### #25 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 07 April 2012 - 08:26

I design aircraft turbine engine drivetrains. We never use tapers or keys for shaft connections. We do use curvics, involute splines, and bolted flanges. The bolted flanges are designed to transmit the full torque load (plus a 1.5 FoS) using friction alone. The curvics are designed to transmit the same torque load with sufficient preload to prevent separation. The splines are sized to transmit the full torque load based on a flank contact stress (P/A) limit of <5 ksi.

Take a look at a typical turbine engine bolted flange joint. They are designed to transmit torques using friction alone, and that's why they use large numbers of bolts.

slider

Im not english so i don`t know what FoS is. "(plus a 1.5 FoS)"

<5.Ksi as well.

### #26 Tony Matthews

Tony Matthews
• Member

• 17,503 posts
• Joined: September 08

Posted 07 April 2012 - 09:31

Safety Factor?

Kilopound/square inch

### #27 bigleagueslider

bigleagueslider
• Member

• 1,106 posts
• Joined: March 11

Posted 09 April 2012 - 03:28

Safety Factor?
Kilopound/square inch

As my Mexican-born, US immigrant father would say when my US university educated Irish-born mother would correct his grammar, "I don't speak English, I speak American".

FoS is "Factor of Safety". It is a factor commonly applied to analytical procedures to compensate for inaccuracies that can result from the simplifications used to expedite the calculations. "ksi" is just shorthand for thousands of pounds per square inch.

Strange as it might sound, 25 years ago when I went through college, all of my class work was done in metric (SI) units. But when I went to work in the US aerospace industry, everything was done in "Imperial units" (ie. inches, pounds, etc.). To this day it still is. Most every other US industry (such as automobiles, consumer products, etc.) now uses SI units, but US aerospace still uses inches and pounds.

I'm happy to work in either system. It's just numbers, and one system is really no better than the other. But like most things, it boils down to majority rule. The ground rules are determined by the major players. That's why the standard currency of global trade is presently the US dollar, crude oil is traded in barrels, gold is traded in ounces, and natural gas is traded in Btus.

Didn't mean to confuse or insult anyone, it's just the way my brain works.

Regards,
slider

### #28 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 09 April 2012 - 09:46

FoS of 1.5 isn`t much is it? whats common to have where i allways wonder. I now there is standards for trains UIC something.. but at school it was allways FoS of about 3 and so on.

Edited by MatsNorway, 09 April 2012 - 09:47.

### #29 Greg Locock

Greg Locock
• Member

• 4,854 posts
• Joined: March 03

Posted 10 April 2012 - 23:15

FoS of 1.5 isn`t much is it? whats common to have where i allways wonder. I now there is standards for trains UIC something.. but at school it was allways FoS of about 3 and so on.

We used to call a FoS a factor of ignorance. If you know everything about your system you don't need one at all. For instance, when back calculating a real world failure you don't use any fudge factors, as they are what you are probably trying to work out. The best we ever did was a system that relaibly hit 90% of the theoretical strength of the indicvidual components, although that was a simple tensile test, and you could almost certainly do better than that with a very simple system.

The same system was designed to work for 400 cycles of deployment and retrieval, that was the tricky bit.

### #30 bigleagueslider

bigleagueslider
• Member

• 1,106 posts
• Joined: March 11

Posted 11 April 2012 - 00:45

FoS of 1.5 isn`t much is it? whats common to have where i allways wonder. I now there is standards for trains UIC something.. but at school it was allways FoS of about 3 and so on.

MatsNorway-

The appropriate FoS depends upon the application and the particular nature of the analysis. The FoS is usually lower when weight is critical. For example, aircraft primary structure designs qualified by analysis would use a 1.5 FoS, while the same designs qualified by both analysis and structural test would use a 1.25 FoS. For extremely weight critical applications like spacecraft, where the payload and structural mass component must be less than 10% of the total gross launch weight, a small change in FoS can have a huge impact.

For applications like lifting equipment or pressure vessels, where weight is not critical but safety is, the FoS by analysis can be 4.0 or 5.0.

You are correct about a 1.5 FoS appearing quite small for most applications. But you must also consider the compound effects of other knockdown factors that are typically applied to the analysis for things such as stress concentrations, load reversals, weld defects, material quality, dynamic effects, etc.

Regards,
slider

### #31 John Brundage

John Brundage
• Member

• 294 posts
• Joined: July 08

Posted 12 April 2012 - 01:51

My brother used to play with train axles a few years back. He explained to me how they would press the old wheels off the axle and then crack test the axles and then fit new wheels. He made it all sound easy.

I talked to the crew that machines the train wheels where I work. The shafts are not keyed. Depending on the composition of the wheels, they usually have a .003" interference. It takes 60-80 tons of pressure to install the wheels on the shafts and sometimes 200 tons to remove. I have pictures of the press that I can post next week. On my way to VIR for the weekend.

### #32 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 12 April 2012 - 07:53

Depending on the composition of the wheels, they usually have a .003" interference. It takes 60-80 tons of pressure to install the wheels on the shafts and sometimes 200 tons to remove. I have pictures of the press that I can post next week. On my way to VIR for the weekend.

For Metric users thats 0.2mm pressfit. We don`t cold press here.. yet.

### #33 Catalina Park

Catalina Park
• Member

• 6,189 posts
• Joined: July 01

Posted 12 April 2012 - 09:28

I talked to the crew that machines the train wheels where I work. The shafts are not keyed. Depending on the composition of the wheels, they usually have a .003" interference. It takes 60-80 tons of pressure to install the wheels on the shafts and sometimes 200 tons to remove. I have pictures of the press that I can post next week. On my way to VIR for the weekend.

I think that was about the same figures that he told me.

For those that are into train wheels take a look at this...

### #34 MatsNorway

MatsNorway
• Member

• 2,208 posts
• Joined: December 09

Posted 12 April 2012 - 12:19

what the hell was that?

Did it chance the the track gauge on the go?

I would not mind a in deepth look at the system.