Dynamic simulation - 3D contact - problem with "noise" - forces on Output Grapher

[Boloz]

Simple mechanism - two parts - a wheel and a kind of a trailer. The trailer is joined to the wheel using a revolution joint and both parts are connected to the ground using spatial and a 3D contact. The revolution joint is driven - given velocity.  I have a problem only with the 3D contact - I tried many different values of parameters (damping, stiffness, friction), but I still have a chart with "noise" - but only on chart of forces. 

[JDMather]

I didn't dig deep into this yet, but first thing I notice is a much higher friction value (1) for the rolling wheel than for the dragging part (0.1).  Is this intended?

[Boloz]

As far as I checked, it's not a crucial parameter, it only influence on "trailing force" but it doesn't change this noise. I run some tests for 0.1 too.

[JDMather]

I gave it a bit more thought and my initial reaction was incorrect.

I will have to dig deeper into the problem to understand the noise in the curve.

[Boloz]

I started with four wheels, articulated vehicle, and I didn't know why I was receiving noise (I also turned on "display" in 3D contact parameters - vectors of forces were changing very quickly because contact points change all the time during rolling.). So I created this simple variant. The problem occurs in case of a driven wheel and a 3D contact. When I tried with a sliding wheel, external force and 3D contact, it was ok as well as a driven wheel but with a rolling joint RI cylinder on plane

[j.palmeL29YX]

---

@Boloz wrote:

... I still have a chart with "noise" - but only on chart of forces. 

---

This is how the 3D contact works: 

When two parts are pressed against each other, they penetrate each other, (In the reality they are deformed). Hereby a counterforce is generated that tries to move the two parts apart again. This force depends on the stiffness, the damping and the value of penetration. (The more they penetrate, the greater the force). 

In your example: the gravitation moves the wheel downwards. It penetrates the ground plate. The counterforce tries to moves the wheel upwards until the counterforce is null. Now the gravitation again moves the wheel downwards and you get an oscillating vertical force. This you can not avoid. 

BTW: If you have a close look at the vertical movement of the wheel axis you also will see this up and down. 

I hope I could explain the "mechanism" of the 3D contact understandably with my bad English. 

(I did not check if you have still other errors in your model. For my testing I created the parts and the assembly completely new from scratch - unfortunately only in Inventor 2020 format). 

[Boloz]

In my opinion it does not explain the problem. As far as I checked a source of this problem is that Inventor simulates 3D contact as a certain number of contact points (attached file). Those points change all the time and each point generate a contact force. Nevertheless a total force (sum of all forces) should be more or less steady-going or at least it shouldn't change like noise. For example when I decrease stiffnes I obtain "waving" force but it's still with noise (attached file). I tried with more precise simulation I also tried to add some dampness in spacial joint between the ground and the wheel and it didn't make it better.

By the way  I have Inventor 2020 too.

[j.palmeL29YX]

... but if you compare the vertical force (red) and the vertical movement of the spatial joint (blue [scaled and adjusted to get both curves comparable])  you see both values oscillate exact with the same frequency. That can not be a fluke 😉 

[Gabriel_Watson]

I think you might be asking too much of the software. I suggest you move this to NASTRAN as a better solver which could reduce directional mesh noise and other issues.

[Boloz]

Maybe your're right. But maybe there is a solution, for example lowpass filter or changing a number of contact points. I found something like filtering in output grapher but I don't now how to use it - it doesn't change enything.

[Boloz]

I obtained some interesting results - I tried to compare 3d and 2d contacts - I attached charts and assemblies with simulations. In the case of 2d contact during collide, I have much higher forces, but I get steady-going values while rolling and, what is more important, the same values.