This is a great question, but I think it is too broad. There is a variety of techniques used to idealize real life (complex) situations. It depends really on what assumptions you are willing to live with, and what modeling tools you have at hand. Ultimately for real life simulations, you want your hand on some commercial MBD (Multi Body Dynamics) software and a lot of time and money to put together a decent model that doesn't blow up or isn't "too chaotic" to be useful.
Besides computer modeling, there is my favorite subject: Symbolic modeling of multiple connected rigid bodies. That is, building all the equations of motion and kinematics for a system of bodies and figuring out what influence each parameter has on the results.
If you are looking at elastically suspended bodies (bodies held together by springs) then you have to make assumptions such as the springs are massless, and the bodies are rigid and you can formulate all the spring forces as a function of the body positions and velocity (for damping).
But in real life (for example with engine valvetrains), the mass and inertia of the springs are critical to the dynamic behavior and so people have developed techniques such as "multi-mass springs" and "quarter beam methods" to include such effects. Real life springs resonate, clash and surge.
The same may apply to systems with thin beams that cannot be considered rigid but have to include the flexibility in the model. Frankly, it is exciting that this is an active area of research so many years after Newton and Euler described the equations of motions for rigid bodies. If you are interested you can do a PhD in rigid body modeling. As I said, this is a very broad and complex subject.