My favorite wording for introducing things that make models not work properly any more in new versions is more like "enworsement".
Are you by any chance using the f77 solver on 2017 and c++ in 2022 ? The latter one does require pretty different settings and I often saw deviations between results.
Easiest example.
Create a simple pendulum example (1 link, 1 revolute, gravity) and run it in A/View with 5s and 50 steps.
Check amplitide of angular movement vs time. It's decreasing though we had no damping.
Change to something like 5000 steps and everything is fine.
You could also tighten error on the 50 step solution to get there.
Using the f77 solver worked fine without any adjustments.
If tuning the parameters did not help, contact MSC for assistance. They love validation models to challenge their solver.
My favorite wording for introducing things that make models not work properly any more in new versions is more like "enworsement".
Are you by any chance using the f77 solver on 2017 and c++ in 2022 ? The latter one does require pretty different settings and I often saw deviations between results.
Easiest example.
Create a simple pendulum example (1 link, 1 revolute, gravity) and run it in A/View with 5s and 50 steps.
Check amplitide of angular movement vs time. It's decreasing though we had no damping.
Change to something like 5000 steps and everything is fine.
You could also tighten error on the 50 step solution to get there.
Using the f77 solver worked fine without any adjustments.
If tuning the parameters did not help, contact MSC for assistance. They love validation models to challenge their solver.
