Model Predictive Control of Industrial Demand Response for Production Lines

Large energy consumers can actively participate in the energy market and Demand-Side Management (DSM) by providing wide-band flexibility, spreading over time scales from seconds to weeks and months. We propose a conceptual formulation and develop a technical solution to autonomous Demand-Response (DR) at small time scale for industrial applications using the model-predictive control (MPC) approach. More precisely, a repeated optimization problem over a finite rolling horizon is considered aiming at maximization of the economic benefits of a fabric, whereby, in particular, service contracted financial revenues are weighted against reduced production. Such algorithms yield a profitable trade-off by online optimal planning of energy consumption vs. production, thereby considering the machine and process dynamics which in the common industrial practice has been traditionally neglected. The proof-of-concept and design of the algorithm are demonstrated in a case-study of centralized control of a piloting production line for beverage production consisting of a single production machine and a pair of feeding/delivery conveyor belts. The efficiency of the algorithmic solution is demonstrated by numerical simulations.