I am trying to build a testbed for a controller and to run EnergyPlus therein for many years - it might be better thought of as a continuous run. I am using several weather file years.

At some point the weather file has to relooped, i.e. when arriving at the end of the last year the testbed restarts from the first day of the first year. As the weather most certainly does not change smoothly from last day last year to first day first year, there could be some irregularities that bias test results. So, my questions are:

1. How much an influence will such a weather "jump" have on the EnergyPlus run (how large could values change, how long is the persistence)?
2. How could the loop of weather years be implemented via EMS?
3. What role does the EnergyPlus state play there?
4. Has anyone done this before or are there any example files in the repository that might give a good hint?

Thanks for any participance in the discussion.