How to model the resistive moment of a ball joint?

[KernOily]

Guys how would you model the resistance of a ball joint in a pipe stress program? I do not want to use a spring rate/stiffness per se since this will create incorrect reactions due to using a Hooke's Law-type analysis. This particular joint (1-1/2" 1500# ANSI) has a pretty high resistive moment so I need to account for it.

The resistance of a ball joint to rotation appears as a resistive moment in the joint that is analagous to the frictional forces developed under a sliding pipe shoe. The manufacturer refers to this as "flex torque". I am scratching my head to figure this out - surely one of you has been up against this in the past. Just trying not to reinvent the wheel here.

One way that occurred to me was that I could apply a fake concentrated moment equal to the joint's flex torque at the node where the joint is located but that will artificially load the pipe at that node where in fact there is no load. This would give me the correct answer for the expansion and operating load cases but it would not be correct for the sustained and occasional cases. I guess I could remove the fake moment for the sustained case only.



Thanks!
Pete

 

[gstark]

74Elsinore - you are indeed looking at non-linear component behavior. First of all, let's be sure of the performance capabilities of the "ball joint". Does it provide for universal angulation about a defined center of rotation ?
Does it transfer torsion, or is it torsionally free about its longitudinal centerline ? Is it capable of axial extension or compression ? The answers to these questions will define whether you are defining a "true" ball joint. The balance of this thread will assume the answers are such that the joint is capable of pure angulation only.

1. The joint has a threshold moment that must exist prior to initiation of angulation. This moment is highly pressure dependent.

2. Once this threshold moment is developed, the joint can be assumed to angulate at a constant spring rate.

3. All joints have specific angulation capability. Your analysis should be predicting local yy and zz angulations, with the RSS value not to be exceeded. Once a joint "bottoms out", additional load transfer can occur.
If angulation capability is exceeded, your solution is highly suspect. And, if you are predicting high angulation values, non-linear geometry effects should be considered.

I don't know what analysis software you are using, but I have seen Nastran techniques to address modeling.

I can go on and on about ball joints, as they are a pricipal design element for high temp pneumatics systems.

Based on your name, are you in fact, enjoying the lake filling up ? If so, I am very close to you, at the other lake to the east.

Regards,

Glenn

 

[KernOily]

Glenn thanks for your reply. My answers are in parentheses.

>Does it provide for universal angulation about a defined center of rotation ? (Yes this joint is free to rotate about all three axes. The max rotation is 33°.)

>Does it transfer torsion, or is it torsionally free about its longitudinal centerline? (No, it will not transfer torsion.)

>Is it capable of axial extension or compression? (Nope)

>The answers to these questions will define whether you are defining a "true" ball joint. (Yes, this is a true ball joint, not a gimbal or a hinge or sliding ball joint.)

The balance of this thread will assume the answers are such that the joint is capable of pure angulation only.

1. The joint has a threshold moment that must exist prior to initiation of angulation. This moment is highly pressure dependent. (Yep, it is 550 ft-lb.)

2. Once this threshold moment is developed, the joint can be assumed to angulate at a constant spring rate. (I don't see how this is the case. It does not follow Hooke's Law, i.e. F=kx won't apply because the resistance is a threshold value, not a linear function of displacement? The resisitance to rotation does not increase linearly as the displacement is increased. Maybe I misunderstand your point.)

>I don't know what analysis software you are using, but I have seen Nastran techniques to address modeling. (CAESAR II is what I am using.)

>Based on your name, are you in fact, enjoying the lake filling up ? If so, I am very close to you, at the other lake to the east. (No, the handle comes from the fact that I am a vintage motorcycle racer. A '74 Honda Elsinore is one of my bikes. I think they named it after the Lake Elsinore GP race that is held there, though.)

Thanks for your help Glenn. Pete


Thanks!
Pete