Small commercial buildings use multiple Roof Top Units to condition their open space. Therefore, it is hard to disseminate the individual effects of indoor and outdoor disturbance factors. This work studies the influence of these factors on energy consumption in a cooling season for energy efficiency practices. Because of the interaction between the units in an open space, an optimal controller needs to be constructed using a reduced order model, that captures the predictions from computational fluid dynamics. The sensitivity of the cooling cost to the variation in the input variables, in a global fashion, is determined using Polynomial Chaos Expansion.