Complex computer simulations do more than mimic nature in producing similar values for measured parameters, they also mimic nature in that their complexity makes them difficult to understand.
Thus it is often valuable to model the situation with very simplistic models that can be understood fully and solved analytically. In this process one accepts that large errors due to the simplifying assumptions are being made, the point here is not accuracy, but understanding. The aim is to model the most important aspects of the system, gain insight, and then verify the details on the full, complicated numerical simulation.
The process was iterative, the simple analytics suggested ways to create the numerical models, the answers of the numerical models posed questions that could be understood by the simple analytical models.