Mats,
I toyed with the idea of using BMX and it's promising pseudo stochastic opportunities.
But I dismissed it because about a year and a half ago I tried using it on a project and very quickly ended up raising cases:
these are SPRs with 'Evaluating' as the status up to Creo4.0
I could have raised more as general usability was poor and I felt as if I was testing the s/w and reverse engineering it's operation. It was fighting me at every turn as I tried to pick ways through to get answers. I think it should be given to companies to stress (in the banking sense and not the engineering sense) on real engineering problems involving simulate to get the product into shape.
Documentation was not extensive, file and directory behaviour foreign and immediate CS unable to help (they cannot be experts in everything!).
So our BMX maintenance lapsed as a result and I would need some significant persuasion to spend more time (and money) on it.
Sad but true.