11 Replies Latest reply on Nov 27, 2012 12:04 PM by Bill McEachern

# Train wheel problem

Hey everyone.

I need some help with a school assignment and I hope someone can help me!

This is the part that I'am doing the FEA analysis on.

At the moment I've concluded that it's to external loads that the trainwheel is exposed too. Torque from the motor and the load from the train wagon.

I've placed these two loads at the like this, But here is the question. How can I use the fixtures so I can get an result that's the most realistic. The fixtures that I've placed at the moment are a fixed geometry at the contact area of the wheel and the railway track(I haved drawn that here). And that's actually everything I have at the moment. I'am opened to everything that might help me.. Kindly regards Martin

• ###### Re: Train wheel problem

Martin,

Good start. You've pretty well captured the free body diagram on a wheel.

Consider the following:

• If the train is turning, there will be another torque around the direction of travel. I don't know what level of physics you have, and that stress is complicated. But mention the assumption that you are going straight.
• Unless this is a driver wheel, which it doesn't look like, the torque around the axle should be negligible.
• The bearing load cannot go 180 degrees. It is a function of the cosine of the angle from vertical, so it goes smoothly to zero at the sides. Solidworks has a bearing fixture, and a bearing load.
• I'm curious how you modelled the contact with the rail. This will be your point of maximum stress, and it is the trickiest to model correctly.
• ###### Re: Train wheel problem

I'am thinking of a scenario were the train are at startup, but I tried too do it without implementing spin and friction. Is this something I need to have?

I did actually think that this wheel was a driver wheel so that's why I brought in torque.

When it comes to the modelling of the contact between the rail and the wheel. I have tried a cut and a splitt line but both of them produces a very high stress concentration at the concers of the contact area.. I assumed it was a rectangular contact area with a contact of 155mm^2.

I did change the force to a bearing load and differance wasn't a big differance thou. The thing is that I'am not that certain of how I use the bearing fixture. By the way I did actually splitt line the contact area of the wheel and "the axle".

Regards Martin

• ###### Re: Train wheel problem

It doesn't suprise me that the bearing didn't make much difference. The way your are doing it is pretty common, but typically people use smaller angles--60 deg or so. Just be aware that if that were the stress you are concerned about, you are getting an approximation.

Choosing a contact area is probelmatic unless you chose it from a good reference, like Roarke. Because, the contact area is in a sense the goal of your analysis. If you knew the contact area, you could compute the stress directly with a pencil. Also, the contact area depends on the material of the rail.

The best way to model the contact between the rail and the wheel is to model the rail and fix it at the bottom. Then model the wheel and fix it so that it can move up and down only. Then apply 'no penetration' global contact and let it rip.

This is going to take some horsepower, because 'no penetration' is a non-linearity which will terribly impact you computation time. Also, however you model it, run it over a few mesh sizes to verify the stress isn't going off to infinity. You have to be carefull applying loads along split lines.

• ###### Re: Train wheel problem

no penetration global contact will not work. You need to specifiy a specific contact. If you want to know a lot of this look up papers by a guy named Joe Kalousek - he is Mr. Rolling Contact for rail wheel interactions.

• ###### Re: Train wheel problem

Bill,

I'm curious why global no penetration won't work. I use it routinely, and it does what it describes.

• ###### Re: Train wheel problem

you need to have a defined are of contact for it to work. In this case you will only have line contact at best. I would actually have a small gap and use suface to surface contact for this problem. Also you need to ensure when you are looking at the results for say Von Mises that you interpret the value correctly for pure shear if you want it to line up with say a hand calc of Hertzian contact stress.

• ###### Re: Train wheel problem

Why not use a virtual wall to model the contact to the rail?

• ###### Re: Train wheel problem

Bill,

I disagree with setting the area. This ignores the properties of the rail and is effectively setting the stress. If you choose line contact, for instance, the stress will certainly converge to infinity. The only way this produces a valuable result is if you actually don't care what the stress is at the contact, for instance, if you were evaluating holes in the wheel.

I also don't think that train wheels fail in sheer. Unless this train is turning, and it is a very complicated problem if it is, the only sheer stress is from the torque. I doubt the train has enough traction to rip the wheel apart in this mode. I think the interesting stresses are the bearing stress and the contact stress.

As Phil pointed out, the virtual wall is probably the cleanest. My version doesn't have it, though, so I'd have to model the rail.

• ###### Re: Train wheel problem

I have modelled wheel/rail contact in Motion Analysis;  is it the same algorithim in FEA?

• ###### Re: Train wheel problem

Hi Mike,

I did not say specify and area, I said specify a contact - surface pairs that will be in contact. The contact area, assuming a sufficiently fine mesh wilthen be resolved as willthe shape of the contact patch and the deformations.

You might want to read up something on the state of stress in Hertzian contact - there is a shear component in the state of stress. See sample contact calc attached.