EnergyPlus chiller load is much lower than measured data

edit

Hello,

I have been calibrating an EnergyPlus model for a very old mid-rise multi-family building. My approach involves modifying the DOE Prototype Mid-rise Apartment Building Model and calibrating it against site-measured data.

While the simulation results show a reasonable match for the whole building load compared with the measured data, the chiller electrical load is much lower than the measured data except between 10 AM and 5 PM. I have adopted the empirical chiller data provided by EnergyPlus, which corresponds to the same model as the on-site chiller. Below figure shows the model calibration results.

My objective is to achieve an acceptable chiller CVRMSE against the measured data.

I would appreciate any thoughts or comments you might have to resolve this issue.

Thank you!