Simulation
The settings for and the performance of the solver are usually dependent on the model. So you need to validate, if subsequent comments are applicable to his model:
■In most dynamic simulations, the C++-solver with the HHT integrator seems to offer the highest performance and stability with the default error tolerance of 1.0e-5. Tighter error tolerance should not cause an important increase in CPU-time.
■The GSTIFF/I3 integrator delivers good results, but may require limiting HMAX and/or more patterns. Limiting HMAX may also be a good practice for the HHT solver.
■A too small number of contact planes may show some numerical noise. A larger number of contact planes is suggested for helical gears. A too large number of contact planes increases CPU-time without substantial increase of result quality.
■Unrealistic contact stiffness for the rigid body contact should be avoided. High numerical stiff leads generally to noisy results and in consequence longer CPU-times. It has to be stated here, that the integrators of Adams are generally efficient and robust.
■An unrealistic high hydrodynamic damping forces the integrator to take small time steps and lead to inaccurate results, as the damping force is dominating the load transfer.