This content has been marked as final. Show 6 replies
Joint friction is a capability that accounts for the geometry of a joint on the frictional resistance to motion. Internally is it still based on forces normal to the surfaces that at contacting, but is takes into account the forces and moments applied to the rigid joint and determines frictional forces due to each component. The type of joint is very important as each type of joint has different geometric/frictional characteristics.
Joint friction adds quite more complexity to the dynamics of the model interms of dertermining if something should move or not (ie overcome the frictional forces). You do need to take care that the friction coefficients and more important, that the joint geometry is correctly defined on the friction tab. If it isn't you can get strange behavior that does not represent reality. The online help does cover an explaination of how joint friction works.
One key assumption of joint friction is that the joint origin is at the center fo the joint geometry. This means that if the joint is actually well offset from the geometry, the forces and moments are going to generate incorrect frictional forces.
If results are jagged, look at increasing the number of simulation frames and increasing the solver accuracy (and/or reducing maximum time step).
It is good to focus on only including friction on the joints where you know it may be a problem, especially to start with. Start with one or two joints and make sure you understand the behavior. Look at the forces and moments on the joint to see how these correlate to the friction. This is good to ensure you have correctly defined the joint characteristics.
I hope this helps.
Thank's Ian, I test now some simple models to understand the behavior.
I have a similar question. I am trying to figure out how COSMOSMotion handles the most basic joint friction problem. I have set up a 1" cube to slide on a plate. I have two planar joints, one between the cube and the plate, and one to keep the cube from rotating (between the side of the cube and the side of the plate). I have defined a coefficient of friction and joint area for the mate between the cube and the plate. I have also disabled gravity, and instead applied a normal force. According to classic physics, F=uN where u is the coefficient and N is the normal force. However, when I run the model, I am getting weird results. If I set u=0.5 and N=10 lbs, then the cube will still move forward even if a very small force of 0.1 lbs is applied. Also, the motion is linear. I believe that if the force is large enough to overcome friction, then the acceleration should be constant which would lead to a non linear position curve.
The reason I am running this model is to try and figure out how COSMOSMotion handles friction. My real problem is that I have a more complicated mechanism where I need to know how much force to apply to one component to overcome friction and cause another component to move. Is COSMOSMotion the right tool for the job?
you can get a description of the friction model in motion by doing a search in the motion help (access it by picking tte help icon in any motion dialog) . Do a search on friction and topics appear. It may help your understanding.
If you are able, please attach the model so I can review it.
Realistically, planar friction is a little arbitrary. 3-d contact with friction is a more appropriate way of representing this. The most common problem as discussed earlier is that the joint location with respect to the cg or force position will induce unexpected results. The program assumes the joint origin is at the middle of the bearing region.
The essense of joint friction is that when there are bending moments on a joint, these actually increase the frictional resistance. Think of a shaft in a tube with a load pulling down on the end of the shaft and the tube is a lot shorter than the shaft . The friction takes place where the surfaces rub. With the load being offset, there is a bending moment and this resolves itself as a force couple. The force couple increases the local normal loads and hence the frictional resistance increases (although the friction coefficient doesn't change or the load). The shorted the tube, the large the force couple and hence joint friction.
This is why it is required to specify joint geometry to account for the radius and length of the tube in this scenario.
On trick you can do to eliminate any bending effects from the joint friction calculations is to specify extremely large geomtry values in the mate definition so that the force couple is insignificant. Then you will get a friction force values inline with just the friction coefficient times the normal force.
Hope this helps explain the capabilities in more detail.
That helps a lot. I had no idea that the joint dimensions were used to calculate bending moments and adjust the friction. I will try a few different things to see if I can get more realistic results, and then I'll probably be back with more questions. Thanks!