A Computationally Efficient, High-Fidelity Testbed for Building Climate Control

Abstract

Advanced building climate control systems have the potential to significantly reduce greenhouse gas emissions and energy costs, but more research is needed to bring these systems to market. A key component of building control research is testing algorithms through simulation. Many high-fidelity simulation testbeds exist, but they tend to be complex and opaque to users. Simpler, more transparent testbeds also exist, but they tend to neglect important nonlinearities and disturbances encountered in practice. In this paper, we develop a simulation testbed that is computationally efficient, transparent and high fidelity. We validate the testbed empirically, then demonstrate its use through the examples of system identification, online state and parameter estimation, and model predictive control (MPC). The testbed is intended to enable rapid, reliable analysis of building control algorithms, thereby accelerating progress toward reducing greenhouse gas emissions at scale. We call the resulting testbed and supporting functions the bldg toolbox, which is free, open source, and available online.