Energy efficient control of a boosting system with multiple variable-speed pumps in parallel

The control objective of a water boosting system equipped with multiple variable-speed pumps in parallel is to minimize the pump system energy consumption by control the number of running pumps and their corresponding speeds in a real-time manner, subject to potential changes of (system head) set-points and operating conditions. After a number of static models for different pump combinations are derived, a number of optimal scheduling algorithms are proposed from a formulated Mixed Integer Non-Linear Program (MINLP) problem. The Branch and Bound method is employed to cope with the considered MINLP problem and the Lagrangian Multipliers method is used to handle the corresponding nonlinear programming within each iteration. In order to cope with potential modeling errors, a feedback control mechanism is introduced into the proposed framework. In case of unknown operating conditions, an identification algorithm is proposed to estimate unknown system coefficients in an online manner. The experimental results show a huge potential to improve the energy efficiency of multi-pump systems using the proposed method and algorithms.