Full-vehicle MBD analysis. Problem with vehicle system oscillating under lateral load during cornering

Alex Mamic

I am trying to do some full vehicle MBD analysis research for my Master degree. 

The object of my study is a Go-Kart model. I've set up the non-wizard kinematic model of a Kart using Altair Driver and now I'm trying to do some tests. When performing Straight Line Acceleration/Braking events - everything is fine. Kart accelerates and brakes just as it should, no problem with oscillation can be observed. The same happens when performing the Slalom event with basic settings. The Kart accelerates and turns just like it should. 

But when I'm trying to perform some events that require steering more than a couple of degrees (for instance, ConstantRadius turning event) - I mean any event with long corners, something incorrect happens with the analysis. 

Firstly, the Solver Log shows ERRMES: 

<< ALTAIR DRIVER :: Steering Controller : Computed predicted path error  > allowed path error of 1 mm.
Proceeding with the best approximation.
Possible Reasons:
1. Vehicle has spun out.
2. Path is not smooth
3. Steering wheel angle has reached saturation

and 

<< ALTAIR DRIVER :: Steering Controller : Computed predicted path error  > allowed path error of 1 mm.
Proceeding with the best approximation.

With basic settings of the event, the analysis crashes immediately after this ERRMES. If to perform some custom event with custom .adf (altair driver file -  the 8-shaped track cornering event - I can share it if needed, but can't attach it to this message) and if to decrease the "LOOK AHEAD TIME" parameter of "FEEDFORWARD TRACTION" from basic 0.5 to 0.1 or lower, the analysis may be prolonged till further crash, but to no good results. 

Secondly, while observing the resulting H3D file some weird oscillation may be spotted during cornering (I'm attaching the gif-file of the cornering results). The attached GIF shows the Kart oscillation in the 15-meter Radius corner with speed just of 1 m/s. The Kart behaves in an incorrect way and I'd like to fix it and do my further research. 

Some details about the Kart: both frame and suspension is rigid. There are no dampers and shocks in this type of vehicle. The only damping mechanism in this system is Tire. I believe that the key to solution is in Tire settings file. Maybe you could give me some advise how to set it up correctly for light vehicle analysis?

I'm attaching: 
.mdl file - MotionView model
.gif - visualisation of vehicle oscillation results
.step - CAD-geometry of Kart frame

How can I fix this issue and perform the Go-Kart cornering anaylsis? Is there anything I'm doing incorrect? If you could provide any advise (especially for tire setup for light vehicle analysis), I'd be really grateful.

Thanks,

Alex

GTT Adam

Alex Mamic said:

Hi Adam,

Once again thanks for your support. 

I did as you have suggested:



Unfortunately, this change didn't put the oscillation away. 

Anything else I should try? Just in case, I attach the new version of my model.  

Thanks, 

Alex

Hi Alex,

I used your model to run the double lane change and the GIF can be seen below.

When I first opened your model, I noticed some peculiarly to the model, so I updated your model to 2023.1 and rebuilt the events. I do not observe oscillation, however certain events may still cause that given the earlier topics as well as the fact that I do not see a driver mass, engine mass, etc. The overall mass of this vehicle is 163.892 kg, where 144 kg is just your wheels. 

For other steering events, like the J-Turn, you can see the vehicle struggles for any grip or balance due to the weight issue:

Lastly, the constant radius event:

 

Before doing a deep analysis of the vehicle performance, I would try to model the complete vehicle first before trying to diagnose any dynamics issues with the vehicle layout. 

I've included the modified model. If you have the ability to update to v2023.1 I would recommend it. 

 

Hope this helps!

Adam Reid

GTT Adam

Hi Alex,

When sharing the model, it is necessary to export the model as a model archive:

File >> Export >> Model >> Archive Model & Create Zip File

This will ensure all externally referenced files, such as CAD and tire files, are included in the model and thus can be simulated for debugging. 

 

In terms of the simulation quitting, yes the simulation will terminate if the vehicle deviates too far off of the path as it assumes something has gone wrong. 

 

For the oscillation, the behaviour does not seem abnormal for all the reasons you outlined: the vehicle is extremely light, completely rigid, and has no suspension. In reality, there would likely be some flex in the frame which may help your scenario slightly, but the tire setup is going to be critical in this model I feel.

• What type of tire models are you using?
• Are they generated by a tire testing facility, or are they default tire files that you have scaled considerably to represent your dimensions? 
• Are you using single point contact or cam contacts?
• Is there a similar physical vehicle that you can confirm does not hop like shown in the model?

 

Hope this helps!

Adam Reid

GTT Adam

Adam Reid_21142 said:

Hi Alex,

When sharing the model, it is necessary to export the model as a model archive:

File >> Export >> Model >> Archive Model & Create Zip File

This will ensure all externally referenced files, such as CAD and tire files, are included in the model and thus can be simulated for debugging. 

 

In terms of the simulation quitting, yes the simulation will terminate if the vehicle deviates too far off of the path as it assumes something has gone wrong. 

 

For the oscillation, the behaviour does not seem abnormal for all the reasons you outlined: the vehicle is extremely light, completely rigid, and has no suspension. In reality, there would likely be some flex in the frame which may help your scenario slightly, but the tire setup is going to be critical in this model I feel.

• What type of tire models are you using?
• Are they generated by a tire testing facility, or are they default tire files that you have scaled considerably to represent your dimensions? 
• Are you using single point contact or cam contacts?
• Is there a similar physical vehicle that you can confirm does not hop like shown in the model?

 

Hope this helps!

Adam Reid

Hi Alex,

I took a look into your model and did find an error: you have duplicate tire models currently active in your model.

Your tire entities:

Show as tire pairs, not single tires:

By fixing this issue, your oscillation may disappear.

 

Hope this helps!

Adam Reid

Alex Mamic

Adam Reid_21142 said:

Hi Alex,

When sharing the model, it is necessary to export the model as a model archive:

File >> Export >> Model >> Archive Model & Create Zip File

This will ensure all externally referenced files, such as CAD and tire files, are included in the model and thus can be simulated for debugging. 

 

In terms of the simulation quitting, yes the simulation will terminate if the vehicle deviates too far off of the path as it assumes something has gone wrong. 

 

For the oscillation, the behaviour does not seem abnormal for all the reasons you outlined: the vehicle is extremely light, completely rigid, and has no suspension. In reality, there would likely be some flex in the frame which may help your scenario slightly, but the tire setup is going to be critical in this model I feel.

• What type of tire models are you using?
• Are they generated by a tire testing facility, or are they default tire files that you have scaled considerably to represent your dimensions? 
• Are you using single point contact or cam contacts?
• Is there a similar physical vehicle that you can confirm does not hop like shown in the model?

 

Hope this helps!

Adam Reid

Hi Adam!

Thanks for your reply. 

Firstly, thanks to your advice, I'm able to share the full model with you. Attaching it to this message. 

Answering your questions: 

- I am using MF-TYRE TNO_wheel_205_60R15.tir from Altair standart model library. The only things I've changed there are the radius, width of the tyre, mass, inertia properties to resemble the real kart tyres. After some unsuccessful tests, I've decreased the inner pressure of tyres 10 times up to 0,22. Thought it might help but it actually didn't. Maybe I need to decrease some other coefficients, yet I've no actual idea which ones. 

- Since they're from standart library I believe they are generated by testing facility. 

- As I remember reading through altair help website, MF-TYRE TNO are single point contact tires. Although at the moment I can't find prove for this information.

- Unfortunately, I haven't got any examples of similar physical type vehicle tests to confirm this. But I still believe that this cornering event is possible to be performed at such low velocities like 1 m/s. 

Anyways, I hope that with fully exported model the situation will become clearer. 

Thanks, 

Alex 

Alex Mamic

Adam Reid_21142 said:

Hi Alex,

I took a look into your model and did find an error: you have duplicate tire models currently active in your model.

Your tire entities:

Show as tire pairs, not single tires:

By fixing this issue, your oscillation may disappear.

 

Hope this helps!

Adam Reid

Thanks. This point has really slipped away from my attention. 

Anyway, I've tried this solution right now. The setup I made for front tires (the same for the left side): 



And rear:



Unfortunately, it didnt' help with the oscillations. But still this is a very spot-on reminder. 

Thank you.

Alex Mamic

Alex Mamic said:

Thanks. This point has really slipped away from my attention. 

Anyway, I've tried this solution right now. The setup I made for front tires (the same for the left side): 



And rear:



Unfortunately, it didnt' help with the oscillations. But still this is a very spot-on reminder. 

Thank you.

P.S. Haven't spotted it on 2nd screenshot: Altair somehow changes the option for the first line - Side - from left to right automatically. I'm changing it manually back to right, close the tab and after I open it again - it's changed back to left.

Maybe it's because of and "Asymetric" option turned on. 

GTT Adam

Alex Mamic said:

P.S. Haven't spotted it on 2nd screenshot: Altair somehow changes the option for the first line - Side - from left to right automatically. I'm changing it manually back to right, close the tab and after I open it again - it's changed back to left.

Maybe it's because of and "Asymetric" option turned on. 

Hi Alex,

You will have to delete the existing tires and build new single tires. You have them created as a pair entity so there are actually two tires on top of each other in the model because you do not have them attached to a body pair and point pair. This time, make sure the 'create pair' option is turned off. 

 

Hope this helps!

Adam Reid

Alex Mamic

Adam Reid_21142 said:

Hi Alex,

You will have to delete the existing tires and build new single tires. You have them created as a pair entity so there are actually two tires on top of each other in the model because you do not have them attached to a body pair and point pair. This time, make sure the 'create pair' option is turned off. 

 

Hope this helps!

Adam Reid

Hi Adam,

Once again thanks for your support. 

I did as you have suggested:



Unfortunately, this change didn't put the oscillation away. 

Anything else I should try? Just in case, I attach the new version of my model.  

Thanks, 

Alex

GTT Adam

Alex Mamic said:

Hi Adam,

Once again thanks for your support. 

I did as you have suggested:



Unfortunately, this change didn't put the oscillation away. 

Anything else I should try? Just in case, I attach the new version of my model.  

Thanks, 

Alex

Hi Alex,

I used your model to run the double lane change and the GIF can be seen below.

When I first opened your model, I noticed some peculiarly to the model, so I updated your model to 2023.1 and rebuilt the events. I do not observe oscillation, however certain events may still cause that given the earlier topics as well as the fact that I do not see a driver mass, engine mass, etc. The overall mass of this vehicle is 163.892 kg, where 144 kg is just your wheels. 

For other steering events, like the J-Turn, you can see the vehicle struggles for any grip or balance due to the weight issue:

Lastly, the constant radius event:

 

Before doing a deep analysis of the vehicle performance, I would try to model the complete vehicle first before trying to diagnose any dynamics issues with the vehicle layout. 

I've included the modified model. If you have the ability to update to v2023.1 I would recommend it. 

 

Hope this helps!

Adam Reid

Alex Mamic

Adam Reid_21142 said:

Hi Alex,

I used your model to run the double lane change and the GIF can be seen below.

When I first opened your model, I noticed some peculiarly to the model, so I updated your model to 2023.1 and rebuilt the events. I do not observe oscillation, however certain events may still cause that given the earlier topics as well as the fact that I do not see a driver mass, engine mass, etc. The overall mass of this vehicle is 163.892 kg, where 144 kg is just your wheels. 

For other steering events, like the J-Turn, you can see the vehicle struggles for any grip or balance due to the weight issue:

Lastly, the constant radius event:

 

Before doing a deep analysis of the vehicle performance, I would try to model the complete vehicle first before trying to diagnose any dynamics issues with the vehicle layout. 

I've included the modified model. If you have the ability to update to v2023.1 I would recommend it. 

 

Hope this helps!

Adam Reid

Hi Adam, 

Thank you very much for your effort helping me with my project. 

Seems like 2023.1 version really makes some difference on the solution. Anyways, now I know where should I direct my research to. 

By the way, have you got the model with the Constant Radius event set up (cannor detect it in the archive you've attached)? Having some issues with setting it up from the standart library, still have some errors during solving. If you've got this model left, it'd really help me.

Once again thanks a lot!

Alex

GTT Adam

Alex Mamic said:

Hi Adam, 

Thank you very much for your effort helping me with my project. 

Seems like 2023.1 version really makes some difference on the solution. Anyways, now I know where should I direct my research to. 

By the way, have you got the model with the Constant Radius event set up (cannor detect it in the archive you've attached)? Having some issues with setting it up from the standart library, still have some errors during solving. If you've got this model left, it'd really help me.

Once again thanks a lot!

Alex

Hi Alex,

Not a problem. You will likely see a difference in these results as I simply created the events using the default speeds and sizing. The numerical results should not differ much if the events were identical. 

The model that was used is attached to the previous post. All events are contained with that archived MDL. 

 

Hope this helps!

Adam Reid