To what extent does the credibility of the soil in two-dimensional representation (for example the definition of soil element as shell element in SAP2000) change when exposed to seismic load?
I'm not sure if I understood your question. As we know, soil modelling is always a hard task and sometimes returns unreliable results. And things get worse if we're trying to simulate its behavior when exposed to seismic loads. However, it's the best tool we have nowadays to understand geothecnical materials, and if used with careful and experienced judgment it will add knowledge for sure.
I would say that the credibility of your results depends a lot on the soil you are modelling (plus other things, like geometry, boundary conditions, etc). The constitutive model has an important part on it too. It is common to use linear elasticity, but you can find a lot of scientific papers referring to the use of more complex models, many times with good results
About SAP2000, although it is not commonly used to simulate soil behavior, I don't see a problem to use it to that end.
Sorry if I don't have a straight answer for your question, this turns out to be almost a philosophical questions for geotechnics.