Coupled Wheel Geometry (Fusion 360)

[SillyOldDufferModerator @sillyoldduffer]

[SillyOldDufferModerator @sillyoldduffer]

I'm confused! I'm not sure if this problem is down to Fusion 360 or me not understanding the mechanics governing what happens when two wheels are coupled together. I think it's me.

I made this model in Fusion 360. The two wheels rotate independenly on their axles except they are joined together by a crank. The model has four rotating joints, just like the real thing.

What could possibly go wrong? Well, turning the model's left wheel anti-clockwise by hand, (or automatically by animating the joint), mostly causes the two wheels to turn together exactly as you would expect. But every so often, perhaps every 20 rotations or so, the crank displaces into this second position and causes the right hand wheel to reverse for half a turn. Then the crank re-aligns and the motion returns to normal, until the next time.

If this happened to a real locomotive the driver would cack his pants and the designer's pencil would be ritually snapped !

Can anyone explain what causes this phenomonon? Assuming it's not a software artifact, how should coupled wheel sets be designed to avoid it?

Dimensions: all revolving joints are 10mm diameter. The axles are 180mm apart; the wheels are 100mm diameter, and the wheel cranks are 35mm from the axle centre.

Thanks,

Dave

 

Edited By SillyOldDuffer on 31/01/2017 15:29:57

[Keith LongParticipant @keithlong89920]

Looks like a perfectly permissible action – what you're missing is the set of coupled wheels on t'other side of the loco chassis with the quartering across the axle.

[JasonBModerator @jasonb]

If you don't want to draw the other side then you could align the crank to the frame then it will always stay horizontal

[JourneymanParticipant @journeyman]

It can happen in real life too, thats why there is a connecting rod on each side 90deg apart. As it is drawn at the position when all the pivots are in line there is nothing to prevent the connecting rod ends moving in the same or opposite direction. I am not sure what makes Fusion 360 choose one way or the other but in real life there is bound to be some clearance no matter how small and it depends whether the clearances line up or are in opposite directions. There that sounds as clear as mud now I have written it down!

John

[Anonymous]

It's inherent in the mathematics of the design. When the connecting rod is aligned with the wheel centres the mechanism is unstable and can flip into either of two stable states, which it will do randomly.

Said driver will not need to change his pants as on a real locomotive there is a second set of wheels and crank which are 90° out to the first thus eliminating the unstable state.

Andrew

[SillyOldDufferModerator @sillyoldduffer]

Thanks chaps! That makes sense.

All the time I've spent looking at steam engines without noticing that the cranks on the other side of a loco are out of phase. Now I wonder what other details have passed me by.

Cheers,

Dave

[SillyOldDufferModerator @sillyoldduffer]

Posted by JasonB on 31/01/2017 15:44:16:

If you don't want to draw the other side then you could align the crank to the frame then it will always stay horizontal

That's a really good idea. Do you happen to know if it's possible when the model is an assembly of parts? In this case the chassis and coupling rod come from different drawings.

I'm off to see if Autodesk have anything on their website.

Dave

[Brian OldfordParticipant @brianoldford70365]

Posted by SillyOldDuffer on 31/01/2017 16:15:03:

Thanks chaps! That makes sense.

All the time I've spent looking at steam engines without noticing that the cranks on the other side of a loco are out of phase. Now I wonder what other details have passed me by.

Cheers,

Dave

In the case of two cylinder or four cylinder engines they will be out of phase by 90deg. Those with three cylinders would be at 120deg. UK convention was almost invariably right hand crank leading.

[SillyOldDufferModerator @sillyoldduffer]

Posted by SillyOldDuffer on 31/01/2017 16:21:30:

Posted by JasonB on 31/01/2017 15:44:16:

If you don't want to draw the other side then you could align the crank to the frame then it will always stay horizontal

That's a really good idea. Do you happen to know if it's possible when the model is an assembly of parts? In this case the chassis and coupling rod come from different drawings.

I'm off to see if Autodesk have anything on their website.

Dave

I can answer my own question. It's done using Modify->Align, applied to link the coupling-rod to the chassis at the assembly level. I'm impressed with Fusion360.

[Neil WyattModerator @neilwyatt]

I suspect that although Fusion models the mechanical motions it doesn't model the momentum that would keep the wheels moving in the same direction.

Neil

[Anonymous]

Posted by Neil Wyatt on 31/01/2017 16:40:43:

I suspect that although Fusion models the mechanical motions it doesn't model the momentum that would keep the wheels moving in the same direction.

Momentum doesn't help as it doesn't change the unstable decision point. It's just more spectacular when the system 'choses' the other state.

Andrew

[SillyOldDufferModerator @sillyoldduffer]

Posted by Neil Wyatt on 31/01/2017 16:40:43:

I suspect that although Fusion models the mechanical motions it doesn't model the momentum that would keep the wheels moving in the same direction.

Neil

I don't think it models springs either. I was looking to model a morse bug key where dots are generated by vibrating a flat spring with a weight on the end. Perhaps I should  try a feature request; they seem receptive to suggestions.

Dave

Edited By SillyOldDuffer on 31/01/2017 17:40:00

[Geoff TheasbyParticipant @geofftheasby]

Revelation! I was wondering about this, even mentioned it in an upcoming Club News, ie, could you use a coupling rod only on one side of a 0-4-0 or 0-6-0? I was going to try it on my new loco and see. I was thinking to avoid quartering. Now I don't need to worry, if 'tis to be done, 'twere well 'tis done quickly. Thanks guys!

As for Morse keys, I'm working on that too, but empirically, not on a computer. If you use a spring so, you should perhaps be thinking of pendulums (Pendula?) and oscillators, not springs, as you need to find the point of resonance. This will help to sustain the vibrating item.

Geoff

[Michael GilliganParticipant @michaelgilligan61133]

Posted by SillyOldDuffer on 31/01/2017 17:39:36:

I was looking to model a morse bug key where dots are generated by vibrating a flat spring with a weight on the end.

.

It's not my thing … and may all be obvious to you, but: Rod Elliott of ESP seems to know a bit about Morse Keys **LINK** http://sound.whsites.net/articles/morse-code.htm

MichaelG.

[MuzzerParticipant @muzzer]

You can't easily model springs in CAD so that they work as springs in dynamic assemblies with slides, pulleys etc as far as I know. It's difficult enough to model them as realistic-looking components. Autodesk Inventor includes spring options such as closed ends (I don't think Solidworks does this) and I think if you can be bothered you can use configurations in Solidworks to model different lengths but it's all a bit of a fudge.

The vibration aspect of co-joined springs and masses CAN be modelled in Fusion in the FEA (simulation) environment but I'm not sure what you are planning to simulate? If it's what I think it is, you are talking about a mass on the end of a blade spring. I think you can model that easily enough to determine its natural frequency but whether this can be used in the motion environment may be another matter.

Murray

And on the matter of locomotive wheels, they are generally encouraged to turn in the same direction by the force of gravity acting on the (shared) rails.

Edited By Muzzer on 31/01/2017 18:31:08

[Mark CParticipant @markc]

Murray,

You certainly can model "dynamic" looking springs in Solidworks (IE. they appear to compress/expand as the attached assembly parts move) and you can use dynamic mates and other advanced features to simulate the spring action/force. None of this is under the heading "basic modelling"….

As for the wheel link; the same happens with Solidworks and is down to the uncertainty of the possible options at the point they invert (as mentioned by others). In Solidworks we would probably add a parallel mate relation with the frame part or the horizontal plane (or create some reference sketch geometry if you wanted to waste time and be a smart arse.

Mark

real wheels also have friction applied to the contact with the rail and the axles are rigid fixed points, all of which will help to ensure they generally both turn together.

Edited By Mark C on 31/01/2017 18:41:12

[Neil WyattModerator @neilwyatt]

Posted by Andrew Johnston on 31/01/2017 17:14:06:

Posted by Neil Wyatt on 31/01/2017 16:40:43:

I suspect that although Fusion models the mechanical motions it doesn't model the momentum that would keep the wheels moving in the same direction.

Momentum doesn't help as it doesn't change the unstable decision point. It's just more spectacular when the system 'choses' the other state.

Andrew

I disagree. the 'decision point' is entirely analogous to dead centre of a piston engine – how often does a running by engine reverse rotation without something happening like a misfire or a hydraulic lock to physically counter the flywheel's rotation. With the coupled wheels things can only go wrong at very low speeds or if some force is braking one of the wheels – this can happen in a loco but wouldn't apply in the simulation.

[Geoff TheasbyParticipant @geofftheasby]

S.O.D. try this: http://www.qsl.net/k5bcq/KEYS/KEYS.html

Geoff

[John Stevenson 1Participant @johnstevenson1]

Idle thought.

How often does a full sized engine, or did . loose a coupling rod and what were the consequences ?

I have seen lorries that have lost one end of the prop shaft. Gearbox end is worse as the shaft continues to be driven whilst the vehicle is in motion, pressing the clutch does nothing.

On one smallish truck with a long prop shaft [ larger trucks tend to have shorter shafts and centre bearings ] it was enough to tip the truck over.

And on a Scania tractor unit it was enough to cost many thousands of pound in repairs as it needed a new chassis, air tanks, piping and very expensive brake valves etc.

[SillyOldDufferModerator @sillyoldduffer]

Posted by Muzzer on 31/01/2017 18:29:43:

…

… but I'm not sure what you are planning to simulate? If it's what I think it is, you are talking about a mass on the end of a blade spring. I think you can model that easily enough to determine its natural frequency but whether this can be used in the motion environment may be another matter.

Murray

…

Yes, that's right – a mass on the end of a blade spring. Here's a picture of a freelance key I built without benefit of science. The period depends on the mass, position on the rod, and the spring tension (I think!). In the picture There are two steel weights on the left, the smaller one was only added experimentally to reduce the dot rate.

The mechanism is semi-automatic. Moving the purple paddle to the left closes a contact and allows a dash to be sent. The length of the dash is controlled by the operator. The left movement also tensions the blade spring so that moving the paddle hard right causes the weighted rod on it's blade spring to vibrate. The vibration causes a contact to be repeatedly made and a stream of dots is sent. Even though it's crudely made this key sends about 20 dots per dash which is more than sufficient to do the job.

The mechanism and layout could be improved in several ways.

Now I have Fusion 360 it's very tempting to have a go at a virtual redesign. Being able to model the vibrating spring and weight assembly would be icing on the cake.

For example, I used a bit of hacksaw blade to make the spring. It worked OK, but the balance between tension and blade length is approximate and needed lots of experimentation. I'm sure it could be done much better with a proper spring and some sums. As I'm not too good with maths, it would be wonderful to do it in software by tweaking parameters from my armchair.

Dave

[Mark CParticipant @markc]

At least on a tractor unit it would have spring brakes so it would fail safe but I expect there might be a big kiss pressed on the windscreen at head height!

Mark

[SillyOldDufferModerator @sillyoldduffer]

Posted by Geoff Theasby on 31/01/2017 19:51:51:

S.O.D. try this: http://www.qsl.net/k5bcq/KEYS/KEYS.html

Geoff

Thanks Geoff. I might have guessed someone else got there first,

Dave

[Mark CParticipant @markc]

Dave, That sounds suspiciously close to a dynamic motion analysis – perhaps Murray can enlighten if this is available in fusion?

If it is not, you can probably get all the information you need from section properties dialogues to allow hand calculations (perhaps in excel, given your comments about not wanting do the maths) to be made.

Mark

[MuzzerParticipant @muzzer]

I know it's not what you are saying but there is a related urban myth that air brakes are fail safe because once there is insufficient pressure for the service brakes to work, the handbrake will come on. But you need the best part of 10 bar for full retardation via the service brake yet you only need ~2 bar to get the hand brake off. That leaves a gap where you may have only minimal braking before the hand brake finally comes on. I can tell you this from experience, having taken a 12t bus down a steep gradient at Millbrook with a failed compressor – quite a stimulating experience.

Murray

[Author]

Posts