Model predictive control strategies for industrial demand response of beverage production lines

Utilizing demand response (DR) in industries reduces the need for expensive utilities like storage or backup plants and renders the electricity market more flexible for industrial sites. This paper proposes an MPC-based approach for such sites to respond online to ancillary service requests and participate in DR by controlling optimally the machine speeds of a given number of production lines. We define centralized and distributed optimization programs to cope with a large number of machines within each line and with ancillary services that require a response within seconds (primary services) to several minutes (secondary services). We demonstrate the proof-of-concept by designing controllers for three beverage production lines consisting of filling machines, labelers, shrink packers, and several conveying belts. The efficiency of the algorithmic solution is illustrated by numerical simulations.