Stoked to see more full suspension questions here! Would love to see photos of the full bike - sounds like an interesting design.
The old swing arm, do you know exactly what material it was? If I had to guess, the failure was more likely due to fatigue stresses instead of pure ultimate strength. Although possible, fatigue stress is notoriously complex to simulate in FEA. Because of this, many larger brands have dedicated test labs to cycle test their designs in real life.
Knowing material, shapes, wall thickness, etc of the failed parts should help give a lower level baseline for the new robust design.
From the screen shot, I see some non-standard components acting on the suspension.
How much does that hub motor weigh? Since the motor is mounted to the end of the moving linkage, your sprung/unsprung mass ratio (and pivot forces accordingly) will be much different than ‘standard’ full suspension bike for trails.
Instead of ‘what is the max load’ question, a better question to ask is what suspension events are you expecting this vehicle to encounter? Are you planning for fully loaded dropping off a curb? Or sweet loading dock sends to flat? Each scenario will require different strengths in the design. A free body diagram would be helpful to start.
I’ve used this calculator in the past to run gut-check numbers for loading: Energy of falling object
“What displacement/stress tolerable” is dependent on exact material choice. This is a much broader engineering question - I would suggest researching material mechanics, material data sheets, and beam bending problems. Here’s a good starting point to dive into: Mechanics of Materials - Engineer4Free: The #1 Source for Free Engineering Tutorials
The FEA simulation does not look representative of the real system and will give an incorrect result. Here’s an excellent write up of the mechanics of FEA and how we can approach it in framebuilding: FEA for Framebuilding
Specifically, there should be an axle of fixed length preventing the swingarm from caving in. Hubs are a significant strength member in the overall system.
A distance or slip constraint should be enough. You could also model in a dummy axle and leave it rigid.
For fixed geometry, I would only fix the main pivot location, not the Horst. The Horst pivot will have a reaction force in the linkage that is dependent on your wheel load at bottom out. Linkage Pro software can calculate this, but this is also where the free body diagram of the whole system is helpful.
FEA is not a be-all-end-all and is only as accurate as your constraint system is to real life.
Hope this helps!