autosizing vs hardsizing HVAC system, which one is better choice?

edit

In the case you describe - modeling an existing building - you should use the actual equipment sizes whenever they are available as that will certainly make the model more representative of the building.

Regarding calibration, that is always challenging for the simple reason that there are a huge number of simulation inputs and usually you are comparing your simulation outputs to a very limited set of measured data. Typically only monthly (or maybe hourly) whole building energy consumption is available, without any sub-metered energy data, or zone temperature data, etc. Or in other words, it is an under-determined problem. This means that there are many potential models that will give you a match to the measured data within the error limits recommended by ASHRAE Guideline 14 and/or the IPMVP, not just one final calibrated model.