hexagon logo
Register Sign in
  • State Not Answered
  • Replies 4 replies
  • Subscribers 13 subscribers
  • Views 118 views
  • Users 0 members are here
Options
Related

Solver Issues with General Bushes. Am I missing any theory?

ilapakurti Venkata Naga
Working in ADAMS/CAR 2017.2 version. I am running a full vehicle simulation on a rough road (Kind of Belgian Pave road) at a constant speed.
 
For the Full vehicle model with Torsion and Control arm suspension bushes are set as "Standard bushes (*.bus)) runs without error with GSTIFF/I3 solver.
But, when I replace the standard bushes with "General Bush definitions", the simulation Locks up at the initial stages of dynamic simulation.
 
The simulation runs without error with GSTIFF/ SI2 solver, but computation time is almost 5 times that of GSTIFF/I3 at 0.01 error.
 
Setting error at 0.1 with GSTIFF/SI2 improves the computation time to 1.2 times of the GSTIFF/I3 solver.
 
The question is -
(I) Am I missing any basic understanding that if we use General bushes GSTIFF/ I3 solver is not to be used?
(II) Ultimately, I am calculating Force at the bush attachment points. Compared attachment loads for 0.01 error & 0.1 error results. . Spikes were observed with 0.01 error which are missing with 0.1 error. In other locations - minor variation in the magnitude was observed. Does anyone else have such observations? Is it OK to go with GSTIFF/SI2 error 0.1 when using General bushes?
Parents
  • 0

    First thing is to check the data for the general bushing. Is there something you missed when changing from a standard bushing?

    In general, I would say that those error tolerances are really loose. With SI2, error=1e-2 might be acceptable. But for I3, I would not rely on anything higher than 1e-3.

    To get going, try to set HINIT. It defaults to Hmax/20. And Hmax defaults to the output step unless explicitly set. For a durability run, I would not use Hmax higher than 2e-3, but it might depend on the speed and therefore the frequency content of your inputs. Too many Adams users are afraid of lowering Hmax, thinking it will take longer to solve. In many cases, the simulation can run faster with a smaller Hmax. Look at the timesteps that the solver actually takes, you can see that in the msg file. Then set hmax such that the solver stays at Hmax most of the time. You will likely get faster simulations and better results. Changing of timestep often leads to spikes in forces and accelerations. You should also verify that the order of the predictor stays at 1 or 2. If you see a lot of higher order predictor steps, that indicates that you probably should lower your Hmax.

Reply Children
No Data
© 2025 Hexagon AB and/or its subsidiaries. Privacy Policy | Cloud Services Agreement