I bet my naval architecture students will like this video!!

This newer video (also from Autodesk Inventor) is somewhat better, and at least covers more features: kzfaq.info/get/bejne/nddxjdmEu57TdKM.html

amazing

Кaк сделaть подвижную опору?

I can belive analysis software? I check version Inventor 2014 & Inventor 2019 get different results

Your mesh is too coarse to properly analyze the bending in the center beam. There need to be two layers of mesh elements, as the neutral axis should have zero stress (plus a minimal amount of tension in your case since you fixed both ends instead of having one frictionless). You can tell the mid-point nodes were on, which gives some hint of the distribution, but it is not accurate. It could have easily been used as a demonstration of when you need to refine the mesh (you even mention checking to see if it is fine enough when it is *literally* the textbook example of it not being), how to do it, and then also look at the singularity at the inside corners where it will never converge. Perhaps even point out the rule of thumb that if your deflection is greater than 10% of the beam thickness that you should question the accuracy of the linear solver and consider non-linear analysis. There is so little information in this video it seems irresponsible. People think that if you run an FEA then that is "the" answer, but it is important to remember the limitations and assumptions inherent in the model. You need to do better than this.