Model Predictive Controls

Model Predictive Controls

Model predictive controls (MPC) can yield significant reductions in energy use and peak demand, enable greater responsiveness and stability of the utility grid as alternative renewable energy sources come on line, and improve occupant comfort and the indoor environmental quality of buildings. It is a common control technique in other fields, particularly in the chemical, automotive, and aerospace industries, and has been shown to be potentially beneficial for many aspects of building controls.

The central idea of model predictive control is that a model of the system is repetitively interrogated to determine the best control strategy given some particular set of weather conditions and other sensor readings or other building or occupant-related information, potentially including weather forecasts. LBNL research and development is focused on developing and evaluating low-cost methods of producing controllers that approximate MPC. A field testbed environment has been established that enables real-time co-simulation and testing of prototype systems under real sun and sky conditions.

MPC shading

Example of how shading position and radiant cooling level can be controlled to minimize energy consumption for lighting and cooling over a 24-hour period, subject to zone temperature constraints, weather forecast, and energy cost minimization goals.  Image: Brian Coffey.