Difference between motion on part vs joint?

I have a question about best practices of adding motion to anassembly in CosmosMotion 2007. In SW (w/o CM), a physicalsimulation motor will be mapped to a part motion using CMIntellimotion Builder. This is useful so that if my client does nothave CM, at least I can show the motion using physical simulation.

I have noticed options exist in CM to add motion to the joints aswell as the parts. It appears that adding motion to joints isalmost the same as adding motion to the part. What is thedifference, and why should I add motion to a part versus motion toa joint? Other than the physical simulation being mapped, does onehave an advantage other the other? Is one easier to trouble shootthan the other?

Also, should hardware be added to linkages, such as pins, shoulderbolts, etc? If not, how do you keep the parts assembled during thesimulation in CM? In CosmosWorks, it is necessary to define contactsets for FEA. Is it the same in CM?

Where can I find a tip sheet / check list of best modelingpractices for CosmosMotion, and ? I have mostly used just physicalsimulation in the past, so I am new to CM?
SolidworksMotion Studies
Comments 1
SF
SF 2008-12-06
Hi Chris,

These are good questions and the training courses should cover someof these topics, but here's the basics.

The key benefit of adding motion on a joint is that it alreadydefines the direction of the motion. It is also limited to applyingmotions only in directions the joint is free to rotate.

Motion on a part is more generic and you can define motion in up to6 different directions (3 rotation, 3 translation). Themotors/actuators from physical simulation are mapped to this typeof motion. Since you have to define a cordinate system to let theproduct know what direction is what, it take a little more time toset up. However people in robotics love this one as they can drivethe end of a robot over a presribed path with specific orientationand determine the required movements at each robot motor to achievethis. This is called inverse kinematics.

The consideration for what geometry to included and the level offidelity varies, but here are some guidelines.
1. As much as possible group parts into sets that should move as arigid system. You can do this as subassemblies in SolidWorks or inCOSMOSMotion by dragging and dropping parts together. The size of amodel and the computation time is heavily dependant on the numberof rigid groups you have. You can have thousands of parts in amodel, yet onlyhave 10-20 moving parts. Motion handles these verywell since it only needs the geometry for mass properties (andattachment/contact data).

2. If you're concerned about true mass, included fasteners etc, butattach them to geometry as mentiong in point 1. Most fasteners arenot locked from rotating, so that is add's lots of complexity tothe calculations. Locking them to other parts prevents this andspeeds up solve time.

3. Try to assembly the model the way you would build it in reallife. This is especially important if you need to start gettingaccurate forces and perhaps stress analysis done on components inyour system. It is very convenient to use any geometry to positionparts with respect to one another, but if is going to be much moreaccurate if you connect parts as they would in real life. Thisgives you in essence the load paths through the system and it willhelp out when dealing with redundancies (see next point).

4. Minimize redundancies wherever possible. Redundancies are wherewe use more than one mate to lock a translation or rotation of apart. The best way to imagine this is if you thing of a simple beamthat is pin supported at both ends. For a load at some distancealong the beam, we can readily calculate the reaction forces at thesupports. However, if you have 3 supports along the beam, youcannot easily determine the loads at each support. This is calledstatically indeterminate. To calculate the support loads, you needto account for the stiffness of the beam. Since COSMOSMotionassumes that everything is perfectly rigid, you have the sameproblem. To solve it, the product reviews all the mates and thensimply eliminates any mates restraints that try to lock movementalready restrained by a mate. Since softrware doesn't understanddesign intent, it may or may not remove the right restraint and youwill get unexpected force results and in some cases unexpectedmovement.

With the motion product, what you put into it, is what you get out.With very little effort, you can get basic information out(movement for given inputs). With more effort (groups parts, moreaccurately mating parts, etc) you can get very accurate dynamicmodels the provide excellent load information for sizing motors,bearings, and for structural analysis.

I hope this gives you a little more to go on in using the product.

Cheers,

Ian