Heating, Ventilation and Air-Conditioning (HVAC) units in commercial buildings account for a huge portion of global energy consumption. There is an ever growing need to optimize the energy consumption of an HVAC system along with a system-of-subsystems entity that must be accurately integrated and controlled by the building automation system to ensure the occupants' comfort with reduced energy consumption. To achieve these goals, it is necessary that accurate models be developed that describe the internal dynamics of the system to employ a satisfactory control architecture. This thesis work aims at provide sufficiently accurate models which are able to estimate the temperature, humidity and Carbon Dioxide dynamics in an occupied room. A simplified linear model which describes the dynamics was developed by reformulating the physical equations into a linear regression format. This was followed by the employment of a suitable identification technique to estimate the physical parameters of the system. The second part of this thesis involves the formulation of a two level control architecture to optimize comfort and energy. In this work we propose a model-based framework to maximize the comfort of the occupants using the Predicted Mean Vote (PMV) index. In particular, the set-point control is based on a predictive controller based on a joint optimization of PMV and energy consumption; the low-level Proportional Integral HVAC controllers are autotuned based on simulations of a thermal model. A simulation based validation via a three room test case is presented: the results show the potential for good temperature tracking with a high degree of comfort while also reducing overall energy consumption.