Hello Rzgar, what I am teaching in the course is the following :

In ANSYS many things are set up by default and integration scheme is implicit Newmark This scheme is converging at all steps, but you have to know on which time range 'something reasonable' is happening /you can miss phenomena with a large time step/

1) First of all YOU NEED to permorm auxiliary MODAL ANALYSIS to determine the first non-zero eigen frequency f1 /lowest!!!/ and, therefore, the highest eigen period T1= 1/f1. Then you know the LARGEST time for 'something dynamical is happenning'. And NOW you can select the time step based on this lowest frequency /highest first period/ say DeltaT= T1/ 20 or T1/ 50.

2) Based on further frequencies from MODAL ANALYSIS you can particular study what is happening on this time range.

3) DO NOT FORGET that in ANSYS Newmark integration scheme is set up 'by default' as 'numerically dissipative' by setting 'gamma' - so energy will dissipate and you cannot even observe the standard verification analytical solutions eg for bars, beam and plate. In order to have it 'energy conservative and not dissipative' you should reset 'alpha and beta' correctly.

4) Eigen frequency /especially higher ones/ are mesh-size sensitive. One must be sure that the mesh size is fine enough. Remeshing and convergence study is preferable for the reliable result

Generally, the time step should be less than 1/(20*fmax) in transient structural by ANSYS, where fmax is the highest frequency interested. 

thank you

please, what you mean by  the highest frequency? and how can I define it in my case, while I have thermal load and I want to define the thermal stresses?

Hello Rzgar, what I am teaching in the course is the following :

In ANSYS many things are set up by default and integration scheme is implicit Newmark This scheme is converging at all steps, but you have to know on which time range 'something reasonable' is happening /you can miss phenomena with a large time step/

1) First of all YOU NEED to permorm auxiliary MODAL ANALYSIS to determine the first non-zero eigen frequency f1 /lowest!!!/ and, therefore, the highest eigen period T1= 1/f1. Then you know the LARGEST time for 'something dynamical is happenning'. And NOW you can select the time step based on this lowest frequency /highest first period/ say DeltaT= T1/ 20 or T1/ 50.

2) Based on further frequencies from MODAL ANALYSIS you can particular study what is happening on this time range.

3) DO NOT FORGET that in ANSYS Newmark integration scheme is set up 'by default' as 'numerically dissipative' by setting 'gamma' - so energy will dissipate and you cannot even observe the standard verification analytical solutions eg for bars, beam and plate. In order to have it 'energy conservative and not dissipative' you should reset 'alpha and beta' correctly.

4) Eigen frequency /especially higher ones/ are mesh-size sensitive. One must be sure that the mesh size is fine enough. Remeshing and convergence study is preferable for the reliable result

the time step should be less than 1/(20*fmax)

For thermal stress analysis, I am coupling Transient Thermal to Transient Structural. In Transient Structural, what is the difference between using the default Analysis Settings and copying the discrete times from Transient Thermal's solution? Thank you very much in advance for any information.