No tire lateral forces in quasi static simulation
When i run a quasi static simulation on a vehicle with a lateral force input applied on the vehicle, the vehicle is freely sliding across the road. The tires don't generate any forces in lateral direction. I was anticipating that the tires would generate frictional resistance inline with the friction value defined in the tire property file. If i run the same as a transient simulation, the tires are generating forces in the lateral direction. Does the tires work differently in quasi static and transient simulation? Are there any flags to be activated to enable the stiction behavior in quasi-static simulation? I have tried with both fiala and TNO tire and observe similar behavior
Answers
-
Tire models in general calculate a lateral force based on a derived slip angle. The slip angle is basically the angle between the position vector of the tire and the velocity vector of the tire. In order to have a velocity, the car model needs to be moving. You should not expect a tire model to generate proper lateral forces in a quasi-static analysis. They can generate the proper vertical load (to achieve static equilibrium).
1 -
Chris, doesn't the tires generate any coulomb frictional reaction akin a block sliding on a surface? If a vehicle is stationary and i want to drag it across the road, is it not possible in a quasi static way? Here no wheel rotation would be involved and vehicle is stationary to start with.
Also, consider this other scenario where one tries to estimate static steering effort where the tire is not rotating. How should it be simulated in this case?0 -
"Chris, doesn't the tires generate any coulomb frictional reaction akin a block sliding on a surface? "
No. I would suggest you spend some time reading up on the theory and math behind the tire models you are using, especially if you want to try using if for a use-case it wasn't expected to handle.
For your other question, Fiala is a very low fidelity tire model. It will not give you anything reasonable for static steering effort, even if it worked. TNO tire is higher fidelity, but in general these tire models will not handle static steering efforts well. The physics behind what is happening with the tire and the road is much different under those conditions as compared to when the vehicle is rolling forward (and generating slip angles). Keep in mind, these tire models are primarily used to evaluate vehicle handling, i.e. moving forward at some velocity.
I think F-tire will handle steering efforts, but it is not using a quasi-static simulation.
I will say, that I'm not as familiar with the TNO tire, so there may be some special settings that enable some specific functionality you are looking for. I would suggest reading the tire model's documentation in detail as your next steps.
1 -
Chris, thanks for the clarifications. I guess the answer to my following question could be a "No", but for sake of record asking them - For the two cases that i mentioned which are dragging vehicle across the road and static steering effort, is quasi static mode the only limitation? Could these two cases be evaluated validly with a transient simulation with slow ramping of loads/efforts?
0 -
Karthikeyan M said:
Chris, thanks for the clarifications. I guess the answer to my following question could be a "No", but for sake of record asking them - For the two cases that i mentioned which are dragging vehicle across the road and static steering effort, is quasi static mode the only limitation? Could these two cases be evaluated validly with a transient simulation with slow ramping of loads/efforts?
Quasi-static isn't the only limitation.
Think of it this way. The tire model is like a robot, that's been trained to bake a cake. It bakes cakes very well. Now you ask the robot to cook an omelet. Both the cake and the omelet are cooked and contain eggs and butter and use heat to cook them. But they are completely different requests, so you cannot expect a good omelet since the robot was never trained how to cook an omelet.
For Fiala for certain: The math behind the Fiala model isn't intended for the kind of analysis you describe, especially the dragging case. You can of course, try it for yourself, but I would have no confidence that the results come anywhere near to what is happening in the real world.
These tire models are built around fitting experimental data to empirical math models. This experimental data is measured on machines that record force and moments generated by the rotating tire moving over a flat belt over a range of slip angles, vertical loads, and camber angles. This data is then fit to these math models. Some of the most famous ones are known as "Magic Formula" or "Pacejka" or "Fiala". So, in the simulation, the tire model is expecting as inputs a normal load, a slip angle, a camber angle, and it will generate Fy, Fz, Mx, My, Mz as outputs.
Since "dragging a vehicle across a road" does not fit within the type of data where the experimental data was collected, you should not have any reason to believe that the tire models themselves will work properly or give reasonable results, especially in situations where the slip angle is at or near zero.
As a possible solution/workaround: This is one situation (for the dragging case), where I might consider replacing the tire models with cylindrical graphics, and then defining contacts between the tire and road. Then you can give it a coefficient of friction, etc. You'll likely have to play with the contact parameters to get something to work successfully.
1 -
Thanks Chris, Appreciate your clarifications greatly!
0