To determine whether the step size is producing an accurate approximation, one can test the approximate error by selecting another step size near to that of the first and then comparing the answers. If the answers are very close, then the step size is likely appropriate. If the answers are significantly different, then the step size should be adjusted.
When the step size is large, the approximation is not sufficiently accurate. Decreasing the step size, increases the accuracy of the approximation. Eventually, however, as the step size is further reduced, a more significant truncation error will occur due to the finite word length associated with the computer. Any further reduction in the step size at this point will produce less accurate results because, the magnitude of the truncations will have a greater impact on the lower values of the reduced step size.
As a quick approximation, the time step should be a fraction of the smallest process time constant.
With that said, a Dt of 10-8 would likely be much too small, unless the time constant is very small. The appropriate step size is related to the time constant in the model and can be verified by trial and error using several step sizes.