Tools for Auto-Calibration of Building Energy Models and Fault Detection

With rising energy costs and finite energy supply, Energy Conservation Measures are becoming increasingly important in the United States. This is particularly true in regard to buildings, which account for 40% of the total energy used in America. Currently, energy use in buildings can be assessed through energy-modeling software during the design process or during post-occupancy, once the building is commissioned. But, these metrics are often autonomous from one-another, and not comprehensive. The overarching goal of this project is to develop High Performance Computing support tools that address energy-efficiency and performance demands in buildings. This is accomplished by creating an executive software platform that monitors and makes decisions based on input from: (i) sensor networks integrated into a physical building; and (ii) energy-modeling simulations run on a computer, working in collaboration. The executive program accurately calibrates the energy model to the physical building, provides fault and damage detection feedback to building managers, and establishes the groundwork for predictive controls to optimize building operation. Phase I research produced a working and viable proof of principle prototype of the system, which has been successfully proven effective in calibration and fault detection problems against a simple and well controlled test chamber. Phase II goals include expansion of the executive program and sensor networks to handle all of the sensor and simulation needs of large scale commercial and industrial buildings. Collaborative efforts with commercial ventures, which require well-calibrated building energy models like Energy Service Corporations and Environmental Design agencies, are already ongoing.