Friction in MBD using motionsolve

Yuvarajs91
Yuvarajs91 New Altair Community Member
edited October 2020 in Community Q&A

Hello Sir,

 

I am working of seat suspension mechanism. I am new to MBD and Motion solve.

I need to select the damper of the system.

In the simulation If I use the Friction parameter I am getting error as below.

 

 

ERROR: At time=6.673E-04 the integrator failed to proceed.

----------- Possible Causes -----------------------------------------------------
(1) The integration has become unstable. Tighten (decrease) integr_tol, h_max,
or both in Param_Transient can help stabilize the integration. If the simulation
contains distinctive phases, use multiple Simulate, each with its
own proper integrator parameter setting, to selectively tighten the tolerance
during the period where instability is encountered.
(2) If the simulation was terminated because stepsize has diminished consistently
below h_min, reduce h_min in Param_Transient to force integration to continue.
(3) Non-physical inertia properties, such as mass=100 Kg and Ixx=Iyy=Izz=1 Kg*mm^2,
or extremely small inertia on a part with an unconstrained degree of freedom.
Make sure the modeling data, in particular the part inertia and the gravity, are
specified in proper units consistent with the units given in Param_Unit element.
(4) Beam, flexible body goes out of linear range, bushing has large rotation along
more than one axis, curve goes out of its interpolation range, higher-pair joint
goes out of the range of U or V, etc. Make sure fundamental modeling assumptions,
such as rigid contact assumption used in Force_Contact, are not violated.
(5) Motion displacement defined using LINSPL, AKISPL in dynamic analysis, or as a
function of model states (DX, VX etc), as well as forces defined as a function of
other forces, can cause hard convergence and integrator failure. Avoid these
modeling practices wherever possible.
-----------------------------------------------------------------

 

 

Please can you guide how can i proceed further. and guide me for the friction in Contact.

Please let me know your comments.

 

Thanks and Regards,

Yuvaraj S

Unable to find an attachment - read this blog

Answers

  • Felipe M Leila
    Felipe M Leila New Altair Community Member
    edited April 2019

    Hello Yuvaraj, 

    Does your model run fine if you deactivate the friction on the joints and contact?

    I would recommend you to run your model by steps to identify the error. First include only the joints, motion, springs, forces entities to represent the overall mechanism, does it work? If yes, then you include friction and tune your model to achieve the desired simulation.

     

    Please try to do that and let us know if you succeeded.

     

    Best regards, 

    Felipe. 

  • Chris Coker_21312
    Chris Coker_21312 New Altair Community Member
    edited April 2019

    Yuvaraj,

     

    Just to add some more detail to Felipe's great advice.

     

    It appears that maybe you are using joint friction, as well as friction on your contacts?  These are both very computationally complex, utilizing a LuGre friction model for joint friction, and in general it is common practice to leave these features turned off, unless we can answer two questions:

     

    1)  Is including this friction absolutely essential to capturing the required physics of my problem?  In most cases, the answer is 'no', or can be achieved using computationally more efficient methods like damping.  Also, in the joint friction models, there are different levels of complexity that can be switched on and off.  Only use the friction features that are required.  For 3D contacts, static friction can add a lot of compute time to your runs.  Ensure you need these feature before activating them.

    2)  Do I have all of the correct input data to populate the friction model correctly?  The friction models required details about the size of the joint, friction properties, etc.  Do not assume that you can simply turn on the joint friction model, and use the default values.  Based on the size of your joints, and units of your model, the default values will usually not be a good starting point.

     

    If the answer to either of these questions is 'no', but you need to damp your model's response to achieve a more physically accurate solution, a better choice is to add damping to the model.   This can be easily accomplished by adding a bushing on top of an existing joint, but don't assign it any stiffness properties.  Just give it some damping, in the desired directions of interest.  This is computationally much more efficient that the LuGre model, and can be used with great success.

     

    It's also curious that you are using the VSTIFF integrator, and not DSTIFF.  Also the integrator is taking a very, very small time step, with your max step size set to 1e-08.  If this is required to get your models to run, there is likely some other issue with your model configuration.

     

    Chris