In a Structural Transient analysis, the Initial Time Step is used to begin iterations to search for dynamic equilibrium.

If it is successful, it will do another increment of the same time step. If it continues to be successful, it may increase the time step size, but it will not exceed the Maximum Time Step.

If it is unsuccessful at finding dynamic equilibrium, it will divide the time step in half (bisection) and try again. In your example, that would be 2.5e-2 s. If it continues to be unsuccessful, it will divide the time step in half but it will not go below the Minimum Time Step. In your example, it would be at 1.25e-2 s after two bisections, but it can't do another bisection it can only try again at the Minimum Time Step. It is better to give smaller Minimum Time Step, such as 1/16 or 1/32 of the Initial Time Step to allow the solution control program to do 4 or 5 bisections while trying to find dynamic equilibrium.