While working as a teaching assistant for introductory experimental physics classes (newtonian mechanics and electromagnetism) for the last couple of years, I faced a certain kind of issue again and again. Of course the course material is hard and some students cope better with it than others. But almost all attendants are struggling with analyzing a given real life problem and modeling it with the tools from the lecture.
Once the formulas are written down, most of them have very little problem arriving at the correct solution, even if it involves some difficult integrals or long calculations. The most extreme case of this was a math student who took the class for her minor. I learned some things about math through discussions with her, and the equations posed no challenge at all. But she just could not graph the forces acting on a rotating point mass correctly or add them together in the right way once they were drawn on the blackboard.
This problem is amplified in exams, where errors creep in because of stress and time. But most students don't even get to the point where they can start calculating things, because the basic premise is often wrong. And because of this they lose a lot of points, as they can't make much progress from a wrong start.
Because we separate the exam into three parts and one of these focuses on "understanding" (very simple one line calculations, the goal is to reason correctly about the physical system and predict its behaviour), we actually have some data on that. In general, this part has by far the lowest mean score while the calculation heavy part fares much better. Also compared to past exams, it seems to have been getting worse, because six or seven years ago this part had a higher mean score.
I am certain, that I had the same problems in the beginning and just learned these things over time through practicing with many examples. But in what way can a teaching instructor or professor teach students to model real life physical systems?
Remark: I chose this site over "Mathematics Educators", because it is strictly about teaching modeling in a physical sense, so I think it fits better here. Also I am not sure, if the "models" tag is appropiate here.
Also if you think this might have something to do with the education system, I am talking about first year bachelor degree courses for physics majors and minors at a German university.
Edit: I will give some examples from memory to clarify the issue.
Example 1 (from the third week): Consider a chairoplane with a given height, rope length and mass of the person inside it. The whole system rotates at a constant angular velocity. Task: model the situation appropriately and draw the acting forces. Problems: a tangential force is drawn, so the person would constantly accelerate in the direction of movement and no force along the rope is considered.
Example 2 (from the 4th week): a ball sits on a table, that stands on the ground, and everything is at rest. Task: draw all forces in the schematic and in the abstraction (that just consists of blocks for ball, ground and table). Problems: no force between the table and the ground is included and forces are not transferred correctly to the abstraction
Example 3(from an exam): a metal rod is lying on a metal loop and there is a current running through everything. The whole system is inside a homogeneous magnetic field and tilted 30 degrees. Task: consider all forces acting on the system and describe its motion. Problems: Lenz's law is not identified to work here; even if it is identified, some ignore it when describing the motion afterwards (so it is described as a rolling motion with the acceleration given by earth's gravity and the slope of the system).