What is Model Predictive Control (MPC) and when should it be used?

MPC uses a dynamic process model to predict future behavior and optimizes manipulated variable trajectory to achieve control objectives while respecting constraints. Benefits over PID: handles multivariable interactions, explicitly handles constraints, optimizes economic objectives, and coordinates multiple manipulated variables. Use when: significant interactions exist between loops, constraint handling is critical, or economic optimization is important. Implementation requires: accurate dynamic model, steady-state optimization (LP/QP), and engineer support. Common in refining, chemicals, and petrochemicals for unit optimization.