Current model-based building control research often uses first-principle or linear identification based building and system model. The detailed high-order EnergyPlus model has rarely been used due to the data interoperability issue and its “black-box” properties. This paper proposes an EnergyPlus Model-based Predictive Control (EPMPC) system to control the under floor air distribution system in a multi-zone open office. The exhaustive search optimization algorithm is utilized in the Matlab/Simulink environment via MLE+ cosimulation tool. Comparing to the baseline control system, by using EPMPC, the HVAC energy consumption can be reduced by 18.9% while maintaining the occupant Predicted Mean Vote (PMV).