question

Have you had a look at the Experimenter? It isn't just for trying different scenarios (though it excels there, too), but you can use it just to gather data across many replications for a single scenario.

It is probably worth working through the experimenter example to get some experience in configuring and using the experimenter (skip the parts regarding Optimization if that doesn't apply to you).

You can then work on your own scenario. You'll get results faster by using the experimenter than you would by running replications yourself because the experimenter is able to run many replications simultaneously, based on the number of cores in your computer. Your performance measures will also be presented with confidence intervals. You should be able to tighten up your confidence by increasing the number of replications. In this way you can figure out how many replications you need in order to hit the confidence that you want. Review the experimenter reference if you need any help configuring performance measures or otherwise setting up your experiment.

All the experimenter links above point to older user manual entries hosted here on Answers, but you can find the same materials for your specific version of FlexSim in the software's user manual, found from the main menu, Help>User Manual. There is a section called Experimenter/Optimizer.

Lastly, if the real-world data for each workstation points to different distributions, you should use the real-world data. If you try to unify them, you will end up with a simulation who's inputs don't match the system you're simulating, and thus the outputs won't be relevant to your real-world system. Stick with the real-world-derived data, unless you have a good reason to change it (like you want to see what a different machine with a different performance profile would do to the system, etc).