Big Data and Model Predictive Control for Improved Building Performance

There has been growing interest in the application of ‘big data’ in recent years to understand and reduce energy usage in a range of areas. However, what are the opportunities to use these data to automatically improve a building’s performance? By using a predictive model of the building to test control strategies in advance of their application, Model Predictive Control has the potential to optimise a buildings energy use and internal conditions. This paper will tackle the question of whether it is possible to develop a suitable predictive model from data taken from a building in operation using the case study of a naturally ventilated school. We show that a building-in-use is rarely perturbed enough to provide adequate data to develop a control capable model.