Dynamic Mechanism Analys - decrease the dynamic effects

[VR_9713318]

Hello.

I'm running a similar analysis, and currently, I'm facing problems to decrease the dynamic effects.

@ptc-4639841, about this Dynamic Analysis that you've mentioned, have you compared the values with the Static condition?

In my model, if I run a Dynamic Analysis, the pin "X" of my 4-bar system will get ~18000 N, but if I use Force Balance, this same pin (in the same equipment position) results in only ~14800 N.

I have run the analysis with 0,1 mm/s using relative tolerance of 1E-5, and friction is turned off.

The accelerations at the fastest point of my mechanism are small. 

Magnitude and mm/s². The peak in the graph occurs when one cylinder stops and the second starts. At the position mentioned before, the graph gives me 0,00075 mm/s².

Sadly, I don't have the BMX license to try James' solution using Multi Objective.

Do you have any idea why, even with these small velocity and accelerations, the difference between dynamic and static is so big?

Thanks!

----

This post was originally posted as a reply on this topic.

[ptc-4639841]

It has been a while, I don't think I was ever able to get the dynamic analysis to work the way I wanted and had to use Multi Objective, which did work quite well. 

Do you have a simplified model you'd want to share? 

[VR_9713318]

Got it. It really does not look easy and reliable to use dynamics for this.

Sadly, I can't share the model. 

I'll keep running some tests in the future, but for now, I'll use the Force Balance and a lot of position servos to get the variations... It's safer do it like this.

I hope PTC can make some improvements in the future, to provide a straightforward way to vary the positions inside MDO.

I'll keep you posted. Thanks!

[ChrisKaswer]

How large are your masses? Most often, I find users miss assigning a material property (density) to one, or a handful, of the components in the system...yes, even the ground part, as you may have this take-on motion at some point as well. Just a thought, but is most often this is the reason you would see these differences.

Good Luck

[VR_9713318]

Hello, Chris.

We are talking about ~750 kg of product and 150 kg of load.

All the masses and loads are correctly assigned, and used for both, Dynamic and Force Balance Analysis.

We are refining the analysis with another model, just to check if this behavior is replicated. We may have something else interfering.

Let's see.

[SweetPeasHub]

My 2 cents... I would want to understand why there is that discontinuity/jump at the peak. It could be some impact event like a joint at its end of travel. Force balance of course would not include the impact. Make sure there are no result changes if you reduce all the mass values to say 1/100 of original Double check all units. Check that friction is truly off. Try with friction very low rather than off. There will be a similar discontinuity with a change in the sign/direction of friction. You probably don't have dampers but they create force by velocity. I have seen issues with very low coefficients of restitution.(below 0.1) The statement about one thing stopping and another starting is hard for me to conceptualize as a 4-bar system.

[VR_9713318]

Hello.

I ran a few tests here, and changed these parameters, but without success.

I went a little bit further now.

The first simulation was done with the complete equipment, which is quite complex, so the differences between dynamic and static were big.

For the second simulation, as a test, I did the following to simplify the mechanism of the equipment:

• create the complete movement of the equipment only with references of sketches inside assemblies. 
• each assembly represent one body. I will call them "body assembly".
• inside each body assembly, manually assembled the heaviest components, to give the right mass and CG position for every body. For the most complex components, created a shrinkwrap or even changed to a solid block to represent them without complexity.

The product was an assembly very clean and light to run.

I know it sounds too much work, but another benefit is that I can reuse the basic assembly with sketches for other models of equipments or in the development iteration (prototype>final product), without the need to reconfigure every measure, motor and analysis again which takes a lot of time. Now I just need to remove the components inside every "body assembly" and replace by the new ones.

Below, the result for one Connection Reaction, represented as a line for the Dynamic Analysis, and as an X for a position with Force Balance Analysis.

The difference is <0,5%. X axys does not matter here, it is just a measure that changes over time and do nor repeat, so I can plot Dynamic and Force Balance in the same graph.

Note: in this second simulation I'm using the real velocity for every motor, so I didn't even need to maipulate it to decrease the difference between analysis.

So, my conclusion is that the complexity of the previous assembly, with several unused locked movements and complex geometries, was interfering somehow in the accuracy of the results.

[achirila · Accepted solution]

The accepted solution on this post can be found here.