Rod Hanger and Friction

Posted by: Edward Klein

Rod Hanger and Friction - 10/24/00 11:47 AM

I've run into an interesting situation today. I am using a +XROD to give me a directional anchor in a piping system. That worked fine.

Then, I added friction to Y supports in the model to make sure my offset legs would actually slide and give me the flexibility I expect.

Now, the friction supports seem reasonable, but the +xrod has gone wacky generating an infinite number in the x direction and very large numbers in the other two, just in the sustained case. To top it off, the loads are not being resisted by other restraints in the model, in that the numbers don't balance out.

I changed the xrod to an x and it works fine, giving me an x force in line with my expectations.

Is there something about nonlinearity that makes fricion and rod hangers incompatible?
Posted by: Richard Ay

Re: Rod Hanger and Friction - 10/24/00 02:07 PM

Friction is not "incompatible" with the large rotation rod models. However, anytime you have non-linear restraints in the model, an iterative solution is necessary. Friction also makes a job non-linear.

The first thing to do with friction is to consider the discussion in the Technical Reference Manual on pages 6-16 and 6-17. The conclusion from this discussion is that once you put friction in a job, you can have multiple equilibrium solutions.

Typically, the place to start addressing friction convergence issues is in the configuration file. Try altering the FRICTION_STIFFNESS parameter. The default value of this parameter is 1e6 lb/in, try reducing this, try 500000, then 250000, then 100000, then 50000, until you achieve a stable, realistic solution. Try not to alter the other friction tolerances.

The problem with the SUStained case is that typically only weight (and pressure) is included. Therefore the load vector consists of loads in only the "Y" degree of freedom. If you don't have adequate horizontal restraint (friction doesn't count as "adequate" because you have to have movement to acquire friction forces), the model is unstable. This can also lead to unrealistic friction solutions.

Richard Ay (COADE, Inc.)