Problems with step size in SIMULINK model

I am having problems in my models where the delay is a multiple of 0.5. It seems that due to rounding arithmetic, I am losing precision and getting erroneous results. For example, I have a propagation delay due to some hold time in my model. When I view the waveforms in scope, I am always having trouble viewing the correct timing information. A particular example is where I am expecting delays along different lines. The delays are 30s and 0.2s, and they have no interdependency so no mutual effects. Where I am expecting the 30s delay, I am getting 10s. I tried to manipulate the step size from 0.2 to auto. In this way, I get the 30s delay correctly. However, the other one with 0.2s delay does not come out to be correct. I get 0s delay i.e no delay. It seems I have a step-size mismatch for each case.

I don't have sufficient knowledge in step-size adjustment and would appreciate if someone could explain what is going on.

N.B. I have noticed that SIMULINK has problems when it comes to 0.5-based calculations. Is it somehow related to my problem too?